The pes of Australovenator wintonensis (Theropoda: Megaraptoridae): analysis of the pedal range of motion and biological restoration

Specimens

The holotype specimens of Australian Age of Dinosaur Fossil (AODF) 604 used in the following analysis include: metatarsals I, II, III & IV; pedal phalanges: I-1 (Fig. S1); I-2 (Fig. S2); II-1 (Fig. S3); II-2 (Fig. S4); II-3 (Fig. S5); III-1 (Fig. S6); III-2 (Fig. S7); III-3 (Fig. S8); III-4 (Fig. S9); IV-1 (Fig. S10); IV-2 (Fig. S11); IV-3 (Fig. S12); IV-4 (Fig. S13); IV-5 (Fig. S14).

The osteological descriptions were adequately described in White et al. (2013a) however their skeletal identifications were not interpreted correctly, therefore we have supplied updated supplementary figures with their correct skeletal identifications and accompanying digital reconstructions. Some of the specimens required restoration of their articular facets due to poor preservation. These include metatarsal IV, pedal phalanges I-1, II-2, IV-1 and IV-2. These specimens were restored digitally by projecting the articular surfaces of the buttressing elements to the damaged surface in Zbrush (see below). The cranial and ventral condyle morphology was manipulated to extend the morphology of the existing bone and where absent references to other well figured allosauroid specimens (Madsen Jr, 1976; Brusatte, Benson & Hutt, 2008) were utilised to replicate the correct morphology.

Software and scanning

All the holotype pes elements of Australovenator wintonensis and the Dromaius novaehollandiae pes were CT scanned at Queensland X-ray, Mackay Mater Hospital, Central-Eastern Queensland using a Philips Brilliance CT 64-slice machine producing 0.9 mm slices. Mimics version 10.01 (Materialise HQ, Leuven, Belgium) was used to create three-dimensional meshes of specimens from the CT scans. The Dromaius pes was also magnetic resonance imaging (MRI) scanned at Queensland X-ray, Mackay Mater Hospital. Many of the holotype pes elements were damaged during diagenesis and required some restoration which was completed digitally using Zbrush 4R7 (Pixologic Inc, Los Angeles, CA, USA). Zbrush 4R7 was also used as a first step process to determine the ROM which was used in conjunction with Rhinoceros 5.0 (Robert McNeel and Associates, Los Angeles, CA, USA).

Ungual and sheath restoration in Australovenator

The extent of the Australovenator sheaths’ were calculated using the formula γ_est = 1.54γ_u described in Glen & Bennett (2007). The estimated sheath angle (γ_est) was calculated from 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 dorsal guide for the sheath was developed using a circle, which was identified as being accurately fitted to the dorsal rim of birds’ claws (Pike & Maitland, 2004; Glen & Bennett, 2007). A perfect circle was dorsally offset to the claw’s rim for sheath thickness in Rhinoceros 5.0. The offset synthesized the dorsal sheath thickness which was progressive until the sheaths tip. Multiple cross-sections of the ungual and the dorsal off-set were created using cutting plains developed within Rhinoceros 5.0. These cutting plains were projected from the centre of the ungual circle and were used to section the ungual at multiple intervals. The cross-sections were used as a guide to draw the outer sheath cross-section outlines. The morphology of the Dromaius sheaths were similar to the underlying bone therefore the assumption is here made that they were also similar in Australovenator. These created an outer sheath rib cage which along with the dorsal off-set guide and the underlying ungual were exported into Zbrush 4R7 to replicate the sheath. The restored sheaths are replicated to a distinct point however wear is quite evident on the Dromaius sheaths. The wear facets were replicated on the Australovenator sheaths from removing the distal portions of the sheaths along an artificial sub straight surface plain that the articulated foot was created on in Zbrush (Fig. 3).

Range of motion measurements

The ROM limitations of both Australovenator and Dromaius were determined with a combination of ‘ex vivo’ manual specimen articulation alongside digital ‘in silico’ articulation (Hammond, Plavcan & Ward, 2016) using both Rhinoceros 5.0 and Zbrush 4R7 (Fig. 4). The maximum soft tissue extension and flexion limits were obtained by bending the digits to their maximum ROM without breakage and securing these positions for CT scanning.

There remains some potential ROM error since the foot of the Dromaius specimen was severed from the rest of the leg, meaning tension in a number of tendons was relaxed, and subsequently increased potential ROM. This highlights the impact soft tissue has on ROM in the pes. The severed limbs were unavoidable for this study due to the complications involved in scanning a complete Dromaius, however scans are planned for the future which will focus on the entire limb ROM.

The soft tissue ROM was simulated by CT scanning a fleshed Dromaius pes poised with the metatarsus in a vertical position simulating a walking stance on a flat substrate. The ROM was determined from this neutral position (0°). This reduced the extension values and increased the flexion capabilities however the overall ROM remained the same. The neutral position was simulated with the Australovenator specimens by superimposing the reconstructed pes over the digitally scaled up Dromaius pes (Table 1).

To arrange the pedal elements to allow for soft tissue extension and flexion in Australovenator a 3-D protractor was created in Rhinoceros 5.0. This was done by using the ‘Arc’ tool to create the specific angle required. Poly lines were drawn to intersect the arc or circles centre and then all of the lines were selected to convert to a ‘pipe’ or 3-D mesh in the form of the required 3-D protractor angle. These protractor angles were imported into Zbrush 4R7, which represented the calculated soft tissue extension and flexion values of the Australovenator specimens from Table 2.

Ethics statement

All necessary permits were obtained for the described study, which complied with all relevant regulations. Permission to excavate the specimens from Elderslie station was obtained from the landholders who donated all specimens to the Australian Age of Dinosaur Museum of Natural History (AAOD). During excavation, each specimen was given a preliminary field number for location and storage purposes. All specimens pertaining to the holotype Australovenator wintonensis are allocated the specimen number AODF604. The specimens are stored in climate controlled conditions at the Australian Age of Dinosaurs Museum 15km east of Winton, Queensland, Australia. The Dromaius pes, which was an offcut from farming emu oil, was purchased from an emu farm for the sole purpose of this analysis. No permits were required as this animal is not endangered and was farmed.

Revised skeletal identification of the Australovenator pes

A review of prior publications (Hocknull et al., 2009; White et al., 2012; White et al., 2013a) revealed a number of discrepancies in the initial phalanx identification of Australovenator.

These include: a small phalanx that was initially described as manual phalanx II-2 (see Figs. 28J–28O in Hocknull et al., 2009) and later reallocated to the position of manual phalanx III-1 (see Figs. 2, 21 in White et al., 2012) is here repositioned to its correct position as a pedal phalanx III-3; a second pedal phalanx described as pedal phalanx III-2 (see Figs. 16, 24 in White et al., 2013a) is repositioned as a left pedal phalanx III-1; the pedal phalanges (left and right) (see Fig. 17 in White et al., 2013a) that were replaced with the newly assigned pedal phalanx III-3 are repositioned as pedal phalanges III-2. The initial misidentification of III-I was based on the poor preservation of the specimen in particular the distal end (see Fig. 15 in White et al., 2013a), which appeared wider than the specimen that was initially identified as pedal phalanx III-2. Two of the unguals were also repositioned. Originally identified as pedal phalanx III-4 (see Fig. 18 in White et al., 2013a), it is now correctly interchanged with pedal phalanx II-3 (see Fig. 14 in White et al., 2013a). A new pedal phalanx recently prepared belongs to pedal phalanx I-1 (Fig. S1). It is poorly preserved and required extensive reconstruction. The ventral half of the proximal articular surface is preserved; however, neither medial nor lateral facets could be distinguished. The main shaft is compressed and shattered. The medial condyle is better preserved than the lateral and possesses a deep fossa. It also possesses portions of its distal surface enabling the entire length to be determined. Based on the remnants of the distal surface it appears that the medial condyle is slightly taller than the lateral condyle however it’s indeterminate which condyle was broader (Measurements in Table S1).

ROM analysis

The ROM of each interphalangeal joint articulates in mostly the plane of extension and flexion however the metatarsophalangeal joint allows for adduction and abduction. The digits of Australovenator converge during flexion and spread apart during hyper-extension which is consistent with most theropods apart from deinonychosaurs (Senter, 2009). The abduction of metatarsal IV is questionable due to the poor preservation of its distal end. The three-dimensional digital articulation of the other elements not only assisted in the specimen’s reconstruction but helped more accurately estimate its ROM capabilities. Some of the Australovenator specimens have poorly preserved articular facets. Without reconstruction their ROM could not have been accurately determined. In these cases the ROM still remains a best estimate only.

The ROM of the Dromaius pes revealed distinct variation with and without the presence of soft tissue. The lowest ROM variation was the flexion of the first phalanx of each digit (II-1, III-1 and IV-1) (Table 2). The largest ROM variation comprised the unguals where the bone ROM far exceeded the soft tissue ROM. There was a small amount of ROM variation between II-1 and II-2 in both flexion and extension. Distinct variation exists in the extension capabilities of II-1, III-2, III-3, IV-1, IV-2, IV-3 and IV-4 with the presence of soft tissue (Table 2). The variation in ROM capabilities with the addition of soft tissue was used to allocate soft tissue ROM to the Australovenator pes (Figs. 2, 5, 6 and Table 2).

Dissection of the Dromaius pes and reconstruction of the Australovenator pes

Herein the soft tissue of Dromaius and hypothetical soft tissue features of the Australoventor pes is described along with their effect on the pedal ROM.

The Dromaius phalanges and the metatarsals possess cartilaginous caps on both their proximal and distal ends. Additionally the ROM of the Dromaius pedal elements were found to be more extensive without cartilage. This is because the proximo-ventral cartilaginous extensions of the phalanges to limit flexion and extension. Interestingly similar findings were also demonstrated in the finger bones of Alligators (see Fig. 2H in Hutson & Hutson, 2015).

The digital flexor muscles consist of intermediate flexors (M. flexor perforans et perforatus digiti II (FPPDII) and M. flexor perforans et perforatus digiti III (FPPDIII)); superficial flexors (M. flexor perforatus digiti II (FPDII), M. flexor perforatus digiti III (FPDIII) and M. flexor perforatus digiti IV (FPD IV)) and deep flexors (M. flexor hallucis longus (FHL) and M. flexor digitorum longus(FDL)) (Getty, 1975; Lamas, Main & Hutchinson, 2014; Patak & Baldwin, 1998) (Figs. 7 and 8). The superficial flexor tendons FPDII and FPDIII insert proximally on the respective digit and are perforated by the corresponding intermediate flexor tendon, whereas the fourth is perforated by only the FDL. The intermediate flexor tendons FPPDII and FPPDIII are perforated by the other respective branches of the FDL. Interestingly the Dromaius dissection revealed that the manual movement of FPDII dictated movement of the digit. Unlike FPDIII and FPDIV, the superficial flexors FPDII was not distally attached to the plantar surface (Fig. 7).

The long digital flexor tendon of the hallux perforates the tendon of a short flexor muscle in the tarsometatarsus. FPDII inserts on the ventral surface of the metatarsophalangeal joint capsule of the second digit. The FPDII houses the FPPDII and the FDL. The latter perforates the superficial flexor resulting in medial and lateral tendons which wrap around from the basal attachment. It anchors cranially posterior to the condyle head of pedal phalanx II-1 and terminates with a cranio-proximal attachment to the ungual on either side of the extensor digitorum longus (EDL). The superficial flexor unifies the digit into a functional unit preventing hyperextension. FPDIII inserts on the metatarsophalangeal joint capsule of the third digit. Medial and lateral extensions of the FPDIII wrap from the ventral attachment proximal to the distal condyles of pedal phalanx III-2 and migrate distally to a cranial attachment on the proximal end of pedal ungual III-4 on either side of the EDL. This prevented over hyperextension of the digit. FPDIII is perforated by the FPPDIII which attach to the proximal portion of pedal phalanx III-3 and the first interphalangeal joint. These two tendons are connected by a fibrous vinculum (FV). The intermediate flexor is perforated by the FDL. It inserts behind the condyles of the phalanges, at the third interphalangeal joint and the proximal portion of the ungual. FPDIV attaches to the distal base of the first phalanx and subdivides into four separate tendons, anchoring on the first, second, third and fourth interphalangeal joints. The first attaches ventrally on the metatarsophalangeal joint. The second inserts on the first interphalangeal joint and attaches latero-ventrally to pedal phalanx IV-3. The third and fourth tendons anchor at each interphalangeal joint and wrap the digit from the ventral to cranial surface attaching to either side of the EDL on the cranial surface of the ungual. FPDIV is perforated by the FDL.

The FPDIV inserts on the plantar surface, proximal of the distal condyles of pedal phalanges IV-2, IV-3, IV- 4, the third and fourth interphalangeal joints and attaches to the proximo-ventral side of the ungual (Figs. 7 and 8).

The EDL trifurcates from a small canal which runs along the cranial surface of the tarsometatarsus supplying single or double tendons to the second, third and fourth digits (Fig. 9). This same feature is hypothesised to have trifurcated from the ascending process of the astragalus in Australovenator (Fig. 10G). It functions as the extensor of the second, third and fourth digits (Getty, 1975; Lamas, Main & Hutchinson, 2014; Patak & Baldwin, 1998). Small fibres of the EDL insert on the cranio-distal ends of the pedal phalanges and the cranio-proximal ends of phalanges and unguals. When the tendon is pulled all of the digits extend simultaneously (Fig. 9).

Digit I is absent in Dromaius therefore a direct muscle comparison with Australovenator was not possible. Therefore Dendragapus obscurus Brooks, 1929 (blue grouse) was used as a biological comparison from Getty (1975). The extensor hallucis longus (EHL) attaches to the cranial surface of the second metatarsal and migrates to the cranial aspects of the first digit, anchoring at the metatarsophalangeal joint of metatarsal 1 and attaching to the proximal end of the terminal ungual. Its function is to extend the first digit. The flexor hallucis brevis (FHB) originates from the medial aspect of metatarsal 1 and attaches to the proximal plantar end of pedal phalanx I-1. It’s perforated by the FHL. A medial and lateral arm of the FHL is suspected to have wrapped from the plantar attachment and migrated dorsally behind the distal condyles of pedal phalanx I-1.

The anchor points were at the inter-ungual joint and attach to the proximo-cranial end of the ungual on either side of the EHL (Getty, 1975; Lamas, Main & Hutchinson, 2014; Patak & Baldwin, 1998) (Figs. 8F–8I).

Abductor digiti II originates on the distal half of the medial aspect of metatarsal II and attaches to the medial surface of pedal phalanx II-1. The adductor digiti II originates on the distal lateral surface of metatarsal II forcing an adduction of digit II. Extensor brevis digiti III (EBDIII) originates on the dorsal surface of metatarsal III and attaches to the cranio-proximal end of pedal phalanx III-1. It assists in the extension of digit III. Extensor brevis digiti IV (EBDIV) originates from the dorsolateral surface of metatarsal IV and migrates between metatarsals III and IV attaching on the proximo-medial side of the pedal phalanx IV-1. It assists in the fourth digit’s extension. Abductor digiti IV originates on the lateral surface of metatarsal IV near the proximal end and migrates along the lateral surface to attach to the lateral surface of pedal phalanx IV-1. It assists in the abduction of the fourth digit and the extension of pedal phalanx IV-1 (Figs. 9 and 10).

In Dromaius the plantar pads are broadest beneath the interphalangeal joints and are thickest towards the proximal part of the foot. Ichnological evidence has indicated that digit IV in theropods possessed four pads (El-Gendy, Derbalah & ElMagd, 2012; Pu et al., 2013; Thulborn, 1990). Additionally, seven large cranial scales exist on pedal digit II and IV, sixteen exist on the third digit and larger cranial scales cover the metatarsus. On the plantar side of the Dromaius metatarsus the scales gradually become larger and more pronounced proximally. The ungual sheaths also have distinct wear facets created from constant contact with the substrate, which is more pronounced on digits III and IV with digit II remaining slightly sharper (Figs. 6 and 11).

Similar podotheca morphological features have also been discovered in the fossilised pes of Concavenator corcovatus (Ortega, Escaso & Sanz, 2010) from the Lower Cretaceous of Cuena, Spain (Cuesta et al., 2015). They share a common ancestor along the Carcharodontosaurian lineage as per a resent phylogeny provided in Cuesta et al. (2015).

This common ancestor is quite basal to these groups however Cuesta et al. (2015) described the actual morphology of the skin impressions as resembling extant cursorial birds which justifies our use of Dromaius and Concavenator to reconstruct the skin.

Concavenator has papillae impressions that are closely spaced and are less than 1 mm wide. They have a slightly negative epirelief and a linear pattern perpendicularly orientated to the long axis of the phalanx (Cuesta et al., 2015). The smallest scales were irregular in shape and predominately on the ventral surface whereas the medium sized scales progress in size on the lateral side of digit IV. The plantar pads revealed the presence of small papillae which Cuesta et al. (2015) identified were similar to ostriches. Remnants of ungual sheaths were also present on digits III and IV with an apparent flat wear facet on the fourth (see Fig. 3C in Cuesta et al., 2015). The Australovenator podotheca was reconstructed based off comparisons made with both Dromaius and Concavenator. The plantar surface was reconstructed with small pronounced rounded papillae. Large hexagon like scales cover the cranial portions of each digit, with six scales covering digits III and IV and sixteen cover the cranial surface of digit III. Wear facets were reconstructed on the unguals of digits III and IV whereas digit II was kept relatively sharp. The sheaths were reconstructed based on the average sheath limits dictated in Glen & Bennett (2007) and the underlying bone morphology (Fig. 12). With the estimated soft tissue ROM, internal biology and outer skin developed, the ROM of the pes can be visually synthesised for future foot print creation (Fig. 13).