A methodology of theropod print replication utilising the pedal reconstruction of Australovenator and a simulated paleo-sediment

Specimens

A complete Australovenator pes which pertains to the holotype specimen AODF 604 was used. The specimen included metatarsals I, II, III & IV; pedal phalanges: I-1; I-2; II-1; II-2; II-3; III-1; III-2; III-3; III-4; IV-1; IV-2; IV-3; IV-4; IV-5 (Fig. 2A). Detailed descriptions of each element is provided in White et al. (2013) and White et al. (2016).

The extant phylogenetic bracket of dinosaurs (EPB) introduced by Witmer (1995) is used as this allows for the inference of traits in extinct animals. In this case crocodylians and birds have provided an ‘in vivo’ tool to determine the effect soft tissue had on the skeletal ROM. We decided to use Dromaius novaehollandiae Latham, 1790 (commonly known as an emu) as our ‘in vivo’ comparison with the pedal morphology of theropods due to this bird seems to be morphologically closer than crocodiles (White et al., 2016).

The Dromaius pes was obtained from an emu farm. It was used to determine its ROM with and without soft tissue. The Dromaius specimens used in this analysis include: the distal end of the tarsometatarsus; pedal phalanges: II-1; II-2; II-3; III-1; III-2; III-3; III-4; IV-1; IV-2; IV-3; IV-4; IV-5.

Each pedal element was computed tomography (CT) scanned and the resulting digital imaging and communications in medicine (DICOM) images were converted into 3-D mesh files using Mimics 10.01 software (Materialise HQ, Leuven, Belgium). These files were imported and rearticulated in Zbrush 4R7 3.0 (Pixologic Inc, Los Angeles, CA, USA) (a graphical design package) and Rhinoceros 5.0 (Robert McNeel and Associates, Los Angeles, CA, USA).

Soft tissue

The Dromaius pes was positioned in flexed, extended and weight bearing positions for both computer tomography (CT) scans and magnetic resonance imaging (MRI). The resulting images were converted into separate soft tissue and bone 3-D mesh files which were viewed in Zbrush 4R7 (3.0) and Rhinoceros 5.0 (Robert McNeel and Associates, Los Angeles, CA, USA). The 3-D Dromaius meshes were scaled to the same proportional size as the Australovenator bone meshes. This provided a guide to digitally articulate the Australovenator bones in the same weight-bearing position as the Dromaius specimen which enabled an ‘in silico’ (Hammond, Plavcan & Ward, 2016) restoration of the Australovenator pes considering both soft tissue attachment points and proportions (See Fig. 11 in White et al., 2016).

Claws

The claws of Australovenator were well preserved and did not require reconstruction however their corresponding sheaths were not preserved. The extent of where the sheath extended past the tip of the claw (γ_est) originating from the base of the flexor tubercle, can be determined using a formulae that was developed by Glen & Bennett (2007) following their examinations of fossilised claws of dinosaurs, Mesozoic birds and extant birds. The estimated sheath angle was determined by measuring the claw angle of the ungual bone (γ_u) and multiplying it by 1.54. The claw angle was determined in Rhinoceros 5.0 from the angle created from the distal limit of the flexor tubercle (B_u) and the ungual tip (T_u). The resulting formulae is γ_est = 1.54γ_u (Glen & Bennett, 2007). Cross-sections created through the Dromaius sheath and claws revealed that the morphology of the sheath closely matched the underlying claw. Therefore we restored the Australovenator sheaths to match the morphology to the underlying bone (Figs. 2A–2E; Table 1).

Range of motion

The ROM of the Australovenator pes was determined from a comparative ROM analysis of the extant cursorial bird Dromaius novaehollandiae Latham, 1790, with and without soft tissue. The variance of ROM with and without soft tissue was applied to the ROM of the Australovenator pes which provided an estimation for the soft tissue ROM (White et al., 2016).

The soft tissue ROM of Dromaius was achieved by bending the digits to their maximum flexion and extension limits and securing them in place for CT scanning. The pes was also poised with the metatarsus in a vertical position simulating a weight bearing phase on a flat substrate to represent a neutral (weight-bearing) position (0°). This reduced the extension values and increased the flexion values however the total ROM capability is the addition of extension and flexion values.

To determine the Dromaius bone ROM, the soft tissue was removed and the bones were CT scanned individually. The separate 3-D meshes were imported into Zbrush 4R7 where they were articulated to their ROM limitations using the Dromaius bones as manual references (see Fig. 5 in White et al., 2016). The variation between the bone and soft-tissue ROM was converted to a percentage which was used to later infer the soft tissue ROM of the Australovenator pes. The extension and flexion ROM limits were multiplied by the ROM variation percentage of the Dromaius pes in order to provide an estimation of the soft-tissue ROM for Australovenator (Fig. 3).

3-D Flexible Australovenator Foot Manufacturing

Australovenator soft-tissue pes was reconstructed in silico in the weight-bearing position (Fig. 3B). The soft-tissue proportions were established by creating 3-D meshes of the Dromaius pes with and without soft tissue. These meshes were rescaled individually or together in both Rhinoceros 5.0 and Zbrush 4R7 so the soft-tissue mesh was scaled along the Dromaius bones. The Dromaius bone meshes were scaled to match the size of the Australovenator bone meshes. The Dromaius soft tissue mesh provided an in silico 3-D outline guide for reconstructing the Australovenator soft tissue. The detailed description of each muscle group was provided in White et al. (2016). The skin texture was restored based on the preserved skin of Concavenator corcovatus (Ortega, Escaso & Sanz, 2010) from the Lower Cretaceous of Cuenca, Spain (Cuesta et al., 2015) (Fig. 2H). They share a common ancestor along the Carcharodontosaurian lineage as per a recent phylogeny provided in Cuesta et al. (2015).

In order to make a physically flexible Australovenator foot, the 3-D mesh was sectioned into smaller parts to fit into a 3-D printer using Rhinoceros 5.0. These ridged solids were assembled to form a life sized Australovenator pes. This solid pes was then moulded using Flexithane 40 and then cast using SRT-30 Silicone. The silicone cured into a flexible rubber to resemble the estimated ROM achievable with soft tissue reported in White et al. (2016) (Fig. 2I). The rubber also allowed slight flattening of the digital pads as what occurs in an actual foot (see Fig. 5 in Gatesy, 2001).

The claws were initially made of the same flexible rubber as the rest of the cast; however this was overcome by coating them with araldite glue, which formed a ridged shell resembling keratin sheaths. We used elastic bands to bring digits II and IV closer to digit three (Fig. 2K). This structure enabled us to simulate a suspended articulation prior to making contact with the substrate. During touch-down the elastic bands were severed and the digits returned to their weight-bearing phase posture. The pes was then rolled forward through the substrate simulating the kick-off phase (Fig. 3C).

Paleo-environment and sediment replication for footprint recreation

The stratigraphic sequence of Lark Quarry was initially interpreted as lacustrine and fluviatile in origin (Thulborn & Wade, 1984). Following the periodic influxes of water, fine cross-bedded sands were deposited. These were covered by clay when the water turbulence subsided. The footprints were impressed in a seam of colour banded claystone which was underlain by cross-bedded sandstones (Thulborn & Wade, 1984; Thulborn, 2017).

The sediment we used to make the Australovenator prints, was mixed in order to simulate the palaeo-environment of DSNM prior to being traversed by dinosaurs. The claystone that houses the footprints was replicated by utilising clay from the dinosaur bone-beds from the surrounding areas. The clay was first dried and then saturated with water. Underlying the claystone was fine-grained cross-bedded sandstones (Thulborn & Wade, 1984; Thulborn, 2017) which were replicated with fine-grained sand. The sediment was layered in a 1 × 1 m sediment box which was lined with plastic. The fine-grained sand formed the base, which was overlain with the highly saturated clay (Fig. 2L).

Photogrammetry

The in situ Lark Quarry ‘bipedal’ prints and the replicated Australovenator prints were digitised in 3-D using Agisoft Photo Scan. A Nikon 810D was used to take the photographs with both a 10–24 mm and 105 mm lens. The 3-D meshes were viewed in both Agisoft and Rhinoceros 5.0. The Lark Quarry tracks were photographed shortly after they were cleaned in early 2016.

Measurements

The measurements that are included here were taken using the software package Rhinoceros 5.0. Scale photographs were taken and imported as a background image. This was used to scale the 3-D mesh of the footprint created in Agisoft Photo Scan so accurate measurements could be taken in Rhinoceros 5.0.

Recreating the prints of Australovenator

The Australovenator footprints were made by the flexible foot simulating the phases of touch-down, weight bearing and kick-off (Fig. 3). Creating the footprints took place during several days where the clay was re-saturated at the beginning of each day and by days end had begun to dry out. Subsequently the prints were formed in varying levels of saturation throughout the day which diversified the data set. The highly saturated clay collapsed inward either shortening or narrowing the digits, whereas lesser saturated clay created a distinct print resembling the flexible foot. Following the creation and photography of each footprint the clay was churned over erasing the print, smoothed over to resemble a flat newly deposited clay ready for the following imprint.

The Australovenator flexible foot was forced into the clay at various angles to simulate movement in: a forward direction, a turn to the left and a turn to the right. The left turn resulted in sediment pushed up between digits III and IV (Figs. 4F and 4G) whereas a turn to the right pushed up sediment between digits II and III (Fig. 4H). When the flexible foot prop was forced into the sediment at an angle to simulate a left or right turn it occasionally created parallel grooves in the sediment formed by the replicated skin papillae (Figs. 5A, 5F). When the foot prop was turned in the sediment the digits and heals created curved parallel striations. The digital pad and heal impressions were a mixture of semi-rounded to hexagon convex impressions. When the foot prop was rolled through the sediment it resulted in some sediment suction which created various sediment suction peaks in the prints. The digital pad impressions showed remarkable similarities to theropod skin impressions that were described in Gatesy (2001) (Figs. 5D, 5E). These included densely packed convex hexagon like depressions of the prints base and regions of parallel grooves, which Gatesy (2001) identified as entry striations created by the skins papillae and as the theropods foot entered the sediment.

The claws and some of the digits occasionally formed tunnels in the sediment when the pes was forced forward. Varying the direction the pes was forced into and rolled out of the sediment was to simulate different movements and direction change of a theropod. Exaggerated heel slides resulted in slightly longer prints (Fig. 4B), prints making a left turn resulted in a shallower digit II impression than digit IV (Figs. 4D, 4F and 4G) whereas during a right turn digit II was deeper (Fig. 4H). The simulated digitigrade footprints created tunnel like features in the substrate with the proximal portion of the footprint tapering towards the digit impressions (Figs. 6H, 6L and 6P).

Australovenator print measurements and morphology

Measurements of nine Australovenator prints (Fig. 4) were averaged and compared with a sample of tracks from DSNM. These included the first five prints from a trackway of 11 prints which were identified as a ‘carnosaur’ in Thulborn & Wade (1984) and an ornithopod in Romilio & Salisbury (2011) here referred to as Trackmaker A (Fig. 7); and two prints from a trackway of 8 prints of a distinctly different bipedal dinosaur (ornithopod in Thulborn & Wade, 1984) here referred to as Trackmaker B (Fig. 8; Table 2). A bivariate analysis was employed to help distinguish the variance between these trackmakers. The parameters used in this analysis include; length (L); width (W); total digit lengths (LII-IV); basal digit lengths (BL2-4); basal digit widths (WBII-IV); middle digit width (WMII-IV). These measurement parameters were adopted from the work of Moratalla, Sanz & Jimenez (1988) who initially employed the method to distinguish between theropod and ornithopod prints. Each parameter was plotted against (L/W) to compare individual parameters with the overall proportional size of the print in question (Fig. 9).

The results of the plots showed that the parameters of Trackmaker A and Australovenator congregated whereas the parameters of Trackmaker B were distinct outliers (Fig. 9; Tables 2–3). The only variation from this was the BLIV which showed a clustering of Trackmaker A & B values. The digit width parameters (WBII-IV; WMII-IV) in this analysis did not reveal any distinct separation between the three print types. The clustering of the Trackmaker A with the Australovenator values indicates that its parameters were closer to the prints of a theropod where as Trackmaker B was distinctly different.

The following aspects were also considered with both print types which included: length verses width of the prints with longer prints generally attributed to theropod and wider prints ornithopod (Lockley, 2009) (Fig. 9B); the shape of the claws (sharp or rounded); overall configuration of the foot print (V-shaped in theropods, U-shaped in ornithopods). Interestingly it was demonstrated by Milàn (2006) that the footprint morphology varies greatly depending on the sediment composition and fluid saturation which was subsequently demonstrated with our Australovenator prints. Some prints exhibited U-shaped configurations characteristic of ornithopods (Figs. 4B, 4C, 4E, 4G, 4I). Additionally sharp claw outlines were not always visible (Figs. 4I, 6H, 6L and 7Q).