1. Introduction
Ichthyosaurs (Ichthyopterygia) were a group of secondarily aquatic amniotes whose fossil record extends from the Lower Triassic (Olenekian) through to the Upper Cretaceous (Cenomanian; McGowan & Motani, Reference McGowan, Motani and Sues2003). Four different body morphotypes seemingly evolved during this 167 Ma evolutionary trajectory (McGowan & Motani, Reference McGowan, Motani and Sues2003: 45–48): the so-called ‘basal plan’, ‘stem plan’ and ‘mixosaurian plan’, all characterized by an elongate eel-like shape and probably undulatory swimming style, and the ‘parvipelvian plan’, which attained a tuna-like or thunniform body shape characteristic of all post-Triassic ichthyosaurs. Thunniform ichthyosaurs were clearly convergent on cetaceans, sharks and teleost fishes (Hildebrand, Reference Hildebrand1974), and were probably adapted for efficient long-distance cruising (Webb, Reference Webb1984) and a pursuit predator lifestyle (Massare, Reference Massare1988; Buchholtz, Reference Buchholtz2001 a). Their swimming capabilities have been inferred from outline shape (Alexander, Reference Alexander1975) and, more explicitly, via hydrodynamic (McGowan, Reference McGowan1992; Massare, Reference Massare, Maddock, Bone and Rayner1994) and mathematical modelling (Motani, Reference Motani2002). However, basic comparative anatomy has pinpointed cetaceans as the closest functional analogues (Buchholtz, Reference Buchholtz2001 a). Indeed, McGowan (Reference McGowan1992) argued that differences in shark tail morphology (specifically, enlargement of the dorsal lobe supporting a cartilaginous skeleton) versus the homocercal tail of ichthyosaurs in which the bony skeleton extends into the ventral lobe imparted significantly different mobility properties, as did the shape of their vertebrae (Massare & Sharkey, Reference Massare and Sharkey2003). In stark contrast, Riess (Reference Riess1986) classified the appendicular skeleton of ichthyosaurs into several locomotory grades focusing on the forelimb rather than the tail as the primary swimming organ. Triassic–Early Jurassic taxa including Shonisaurus Camp, Reference Camp1976 and Eurhinosaurus Abel, Reference Abel1909 were designated as the ‘Neoceratodus-type’, with a paddle-like swimming style similar to that of the living Australian lungfish (Neoceratodus). Jurassic–Cretaceous Stenopterygius Jaekel, Reference Jaekel1904, Ophthalmosaurus Seeley, Reference Seeley1874, Aegirosaurus Bardet & Fernández, Reference Bardet and Fernández2000 (Macropterygius sensu Riess Reference Riess1986) and Platypterygius von Huene, Reference von Huene1922 on the other hand constituted an ‘Inia-type’ in which forelimb motions resembling those of Inia (the Amazon River dolphin) were used for primary propulsion. Lastly, the earliest Jurassic Leptonectes McGowan, Reference McGowan1996 (Leptopterygius von Huene, Reference von Huene1922 sensu Riess, Reference Riess1986) and Middle–Late Triassic Mixosaurus Baur, Reference Baur1887 formed the intermediate ‘Leptopterygius-’ and ‘Mixosaurus-types’, respectively. Paradoxically, however, the swimming capabilities observed in extant Inia conflict with the model envisaged by Riess (Reference Riess1986): Inia actually uses its caudal fin as the main organ for sustained swimming, with the forelimbs alternatively providing manoeuvrability within its complex riverine habitat (Klima, Oelschläger & Wünsch, Reference Klima, Oelschläger and Wünsch1980). Such ‘decoupled’ subaqueous locomotion (sensu Blake, Reference Blake2004) has been reported elsewhere in fishes (Webb & Keyes, Reference Webb and Keyes1981) and also specifically advocated by others (e.g. Wade, Reference Wade1984) for the stratigraphically last-known ichthyosaur genus, Platypterygius.
Platypterygius is atypical among ichthyosaurs (Motani, Reference Motani1999) in possessing markedly robust propodials and broad paddle-like limbs that exhibit extreme hyperphalangy and hyperdactyly (Maisch & Matzke, Reference Maisch and Matzke2000; McGowan & Motani, Reference McGowan, Motani and Sues2003). Von Huene (Reference von Huene1923) first remarked on this unusual trait combination, and also suggested that Platypterygius must have possessed poor caudal propulsion because its tailfin was exceptionally small. McGowan (Reference McGowan1972) reiterated this conclusion based on comparisons with Jurassic ichthyosaur taxa; such a scenario requires a significant reversal in long-term evolutionary trending, and is confounded by the absence of any articulated distal caudal series for Platypterygius spp. with which to test the hypothesis. Similar difficulties have likewise faced counter arguments including Thewissen & Taylor (Reference Thewissen, Taylor and Hall2007) who considered forelimb swimming to be improbable in Platypterygius, but failed to produce contrary supporting evidence.
To dispel these lingering inconsistencies we have selected the most completely known species of Platypterygius, P. australis (M'Coy, Reference M'Coy1867) from the Early Cretaceous (late Albian) of Australia (see Wade, Reference Wade1984, Reference Wade1990; Kear, Reference Kear2003, Reference Kear2005; Zammit, Reference Zammit2010 for stratigraphical information and diagnoses), to serve as a model for undertaking detailed osteological comparisons with marine mammal analogues. Our objective was to identify specific skeletal structures indicative of propulsive mode with: (1) a dolphin (Tursiops aduncus), constituting a proxy for fast tail-driven swimming; (2) the dugong (Dugong dugon), representing a slow, tail-driven swimmer; (3) a sea lion (Neophoca cinerea), as a forelimb propulsor; and (4) two genera of phocid seals (Lobodon carcinophagus and Hydrurga leptonyx) corresponding to hind limb propulsors. Note that although pinnipeds retain the capacity for terrestrial movement, their skeletal morphology does not manifest the extreme specializations indicative of ‘underwater flight’ found in sea turtles and penguins (e.g. fusion of the vertebral column to the shell or development of the synsacrum and pygostyle, respectively). In our opinion therefore, they still provide the closest feasible comparator for the original ‘Inia-type’ concept of Riess (Reference Riess1986).
2. Institutional abbreviations
AM – Australian Museum, Sydney, New South Wales, Australia; KKM – Kronosaurus Korner Museum, Richmond, Queensland, Australia; QM – Queensland Museum, Brisbane, Queensland, Australia; SAM – South Australian Museum, Adelaide, South Australia, Australia.
3. Materials and methods
All of our sampled Platypterygius fossils were derived from the Lower Cretaceous (upper Albian) Toolebuc and Allaru formations of central–northern Queensland, Australia, including nine apparently osteologically mature individuals: KKM R519 (Marathon Station, near Richmond, Allaru Formation), 44 caudal centra, complete left hind limb, right femur, partial right hind limb; QM F2299 (Ashgrove Station, Brixton, Allaru Formation), partial coracoid; QM F2453 (Telemon Lease, Dunluce Station, near Hughenden, Toolebuc Formation), complete skull, 83 vertebral centra, partial coracoids, scapulae, humeri, partial forelimbs; QM F2573 (Lydia Downs Station, Nelia, Toolebuc Formation), left humerus, partial forelimb; QM F3348 (Flinders River, Steward Park near Richmond, Allaru Formation), scapula, humerus, partial forelimb; QM F10686 (branch of Flinders River between Flinders and Borree Park Homestead, west of Richmond, Allaru Formation), numerous highly distorted vertebral centra, pectoral girdle, two incomplete forelimbs; QM F12314 (Kilterry Station, north of Flinders River, near Julia Creek, Toolebuc Formation), incomplete coracoid, partial scapula; QM F18307 (Dunluce Station, near Hughenden, Toolebuc Formation), partial scapulae; QM F18906 (Marathon Station, near Richmond, Allaru Formation), partial hind limbs; QM F40821 (Warra Station, Boulia, Toolebuc Formation), fused ischiopubis and possible clavicle element; QM F40822 (Kilterry Station, north of Flinders River, near Julia Creek, Toolebuc Formation), complete right and incomplete left coracoid; QM F40823 (Kilterry Station, north of Flinders River, near Julia Creek, Toolebuc Formation), incomplete coracoid; SAM P44323 (Warra Station, Boulia, Toolebuc Formation), unprepared material comprising skull, articulated series of cervical to dorsal/sacral vertebrae, pectoral girdle elements, forelimb elements.
The comparative morphological sample of extant marine mammal analogues included 39 Tursiops aduncus (SAM M16261, SAM M16265, SAM M16248, SAM M16972, SAM M18095, SAM M19952, SAM M19953, SAM M19967, SAM M19973, SAM M19978, SAM M20733, SAM M20734, SAM M20748, SAM M20877, SAM M21023, SAM M21027, SAM M21231, SAM M21235, SAM M21237, SAM M21242, SAM M21243, SAM M21314, SAM M22409, SAM M22414, SAM M22418, SAM M22441, SAM M22531, SAM M22532, SAM M22549, SAM M23322, SAM M23323, SAM M23324, SAM M23356, SAM M23357, SAM M23367, SAM M23667, SAM M23668, SAM M23669, SAM M23670); 6 Dugong dugon (QM ‘D’, QM ‘E’, QM J4014, QM J21709, QM JM11148, SAM M847); 23 Neophoca cinerea (SAM M11636, SAM M15964, SAM M18234, SAM M19801, SAM 21282, SAM M21500, SAM M21503, SAM M21507, SAM M21543, SAM M21544, SAM M21683, SAM M21686, SAM M21687, SAM M21703, SAM M21704, SAM M21710, SAM M21711, SAM M21714, SAM M21792, SAM M21793, SAM M22082, SAM M22083, SAM M22084); and 3 phocids, including one Lobodon carcinophagus (SAM M10634) and two Hydrurga leptonyx (SAM M4994/001, SAM M16338). Tursiops aduncus was deemed a suitable model taxon within Cetacea because the group is generally conservative in its swimming styles (Buchholtz, Reference Buchholtz2001 b) and, pragmatically, because substantial numbers of complete T. aduncus skeletons were available for study.
Measurements were made using dial calipers (accuracy 0.02 mm), except for the largest elements which required the use of a tape measure (accuracy 1 mm). Vertebral centrum proportions were taken as maximum length, width and height (Fig. S1, online Supplementary Material available at http://journals.cambridge.org/geo). Maximum dimensions of the scapulae followed the long axis of the forelimb (length) or craniocaudal length of the body (width) (Fig. S2, online Supplementary Material available at http://journals.cambridge.org/geo). Coracoid length was deemed parallel to the medial inter-coracoid facet and its width as the maximum perpendicular dimension. All limb components incorporated length as the maximum proximodistal dimension, with width measured from cranial to caudal and depth from external to internal (equivalent to lateral to medial) (Fig. S3, online Supplementary Material available at http://journals.cambridge.org/geo). Three different width measurements were made to capture the shape of individual bones.
4. Comparative skeletal observations of Platypterygius versus extant marine mammals
4.a. Vertebral column
4.a.1. Platypterygius
As in other thunnosaurian ichthyosaurs, all of the centra in P. australis are disc-like in shape and bear amphicoelous articular surfaces. The position and number of rib facets identify the cervical, dorsal and sacro-caudal regions (compare Figs 1e, 2e and 3e; McGowan & Motani, Reference McGowan, Motani and Sues2003) and the ilium has no obvious contact with the axial skeleton (see Druckenmiller et al. Reference Druckenmiller, Hurum, Knutsen and Nakrem2012 for discussion). The atlas and axis are fully fused (apparently from early ontogeny; Kear & Zammit, Reference Kear and Zammit2013), forming a heart-shaped atlanteal facet and subcircular axial facet; there is no indication of centrum co-ossification elsewhere in the vertebral column. A single ventrolateral rib facet occurs at vertebra 47, indicating the position of the dorsal–sacral transition. The anterior and posterior centrum faces remain amphicoelous, but become shallower posteriorly. Post-flexural caudal centra are laterally compressed and somewhat spool-shaped, facilitating lateral mobility (Fig. 4e).
Cervical vertebrae. (a) Tursiops aduncus SAM M21243; (b) Dugong dugon SAM M847; (c) Neophoca cinerea SAM M15964; (d) Hydrurga leptonyx SAM M16638; and (e) Platypterygius australis QM F2453. Mammalian taxa are shown in lateral (left) and anterior (right) views. P. australis is shown in lateral view only. The fifth vertebra (C5) is shown for the extant mammalian taxa, while vertebrae 3–8 are shown for P. australis. The traditional vertebral regions (e.g. cervical, thoracic) are difficult to define in the P. australis vertebral column, and are conventionally numbered according to their position in the vertebral column rather than their position in a given vertebral region. However, the vertebrae of extant mammalian taxa are conventionally numbered according to their position in a vertebral region. For the mammalian specimens used in this study, vertebral number in this and all proceeding figures is therefore given as both the position within the vertebral column (to compare with P. australis) and as the more conventional method of position within a region of the vertebral column. Scale bar is 10 cm across.
Thoracic vertebrate. (a) Tursiops aduncus SAM M21243; (b) Dugong dugon SAM M847; (c) Neophoca cinerea SAM M15964; (d) Hydrurga leptyonyx SAM M16638; and (e) Platypterygius australis QM F2453. Mammalian taxa are shown in lateral (left) and anterior (right) views. P. australis is shown in lateral view only. The vertebrae shown are as follows (left to right): 10th (T3) and 15th (T8) for Tursiops aduncus; 10th (T3), 15th (T8), 20th (T13) and 25th (T18) for Dugong dugon; 10th (T3), 15th (T8) and 20th (T13) for Neophoca cinerea and Hydrurga leptyonyx; and vertebrate 14–23 for P. australis. Scale bar is 10 cm across.
Lumbar and sacral vertebrae. (a) Tursiops aduncus SAM M21243; (b) Dugong dugon SAM M847; (c) Neophoca cinerea SAM M15964; (d) Hydrurga leptonyx SAM M16638; and (e) lumbosacral vertebrae of Platypterygius australis QM F2453. Mammalian taxa are shown in lateral (left) and anterior (right) views. P. australis is shown in lateral view only. The vertebrae shown are as follows (left to right): 20th (L1), 25th (L6), 30th (L11) and 35th (S5) for Tursiops aduncus; 30th (L5) for Dugong dugon; 25th (L3) and 28–30th (S1–3) for Neophoca cinerea and Hydrurga leptonyx; and vertebrae 36–37, 45–47 and 48–51 for P. australis. Scale bar is 10 cm across.
Caudal vertebrae. (a) Tursiops aduncus SAM M21243; (b) Dugong dugon SAM M847; (c) Neophoca cinerea SAM M15964; (d) Hydrurga leptonyx SAM M16638; and (e) Platypterygius australis KKM R519. Mammalian taxa are shown in lateral (left) and anterior (right) views. P. australis is shown in lateral view only. The vertebrae shown are as follows (left to right): 40th (Ca4), 45th (Ca9), 50th (Ca14), 55th (Ca19) and 60th (Ca24) for Tursiops aduncus; 35th (Ca3), 40th (Ca8), 45th (Ca13) and 50th (Ca18) for Dugong dugon; and 35th (Ca5) and 40th (Ca10) for Neophoca cinerea and Hydrurga leptonyx. The position of the vertebrae for P. australis is unknown as the anterior section of the skeleton is not preserved. Scale bar is 10 cm across.
4.a.2. Caudally propulsive marine mammals
Short disc-like cervical centra are common to both T. aduncus and D. dugon (Fig. 1a, b) and all centra are platycoelous. The cervical region of T. aduncus exhibits centrum fusion but this is intraspecifically variable, occurring either within the atlas-axis complex only or additionally throughout the series. In contrast, the cervical region of D. dugon was unfused. Restricted movement between the thoracic vertebrae is evidenced by the spinous processes and prominent rib facets (Fig. 2a, b) which limit rotational (i.e. twisting) movement (Buchholtz & Schur, Reference Buchholtz and Schur2004). Progressive shortening and increased caudal inclination of the spinous process occur in both T. aduncus and D. dugon (Fig. 3a, b) and are associated with enhanced mobility; the point at which this occurs differs among cetacean taxa and is dependent upon the length of the vertebral column involved in propulsion (Buchholtz & Schur, Reference Buchholtz and Schur2004). Structural homogeneity (i.e. absence of obvious boundaries between the lumbar, sacral and caudal segments) was observed in the vertebral columns of T. aduncus and D. dugon, and is otherwise recognized as a typical feature of cetaceans (Thewissen et al. Reference Thewissen, Cohn, Stevens, Bajpai, Heyning and Horton2006). The caudal vertebrae lack spinous processes and have dorsoventrally compressed centra that facilitate vertical movement (Fig. 4a, b) in the most mobile section of the vertebral column.
4.a.3. Pectorally propulsive marine mammals
Cylindrical, unfused, platycoelous cervical centra occur in N. cinerea (Fig. 1c). The spinous processes show a distinct increase in height posteriorly and are tallest in the thoracic region (Fig. 2c). The posterior thoracic zygapophyses interlock tightly (Howell, Reference Howell1929), limiting movement between adjacent vertebrae. The rib attachments extend over consecutive vertebral arches, a feature which is clearly associated with enhanced rigidity (Buchholtz & Schur, Reference Buchholtz and Schur2004). The lumbar, sacral and caudal segments are readily distinguishable, although the lumbar vertebrae are morphologically very similar to the posterior-most thoracic vertebrae except in the absence of rib facets (Figs 3c, 4c). The lumbar centra are cylindrical (length > width > height), and there is a fused sacrum comprising three vertebrae as well as the first caudal in some individuals (e.g. SAM M11636, SAM M19801, SAM M21503, SAM M21507, SAM M21544, SAM M21683, SAM M21710, SAM M22082, SAM M22083 and SAM M22084). The sacral centra progressively decrease in width posteriorly. The spinous processes also become shorter through the caudal series, and are virtually absent in the last few vertebrae. The caudal centra are generally cylindrical with circular articular surfaces (Fig. 4c).
4.a.4. Pelvically propulsive marine mammals
Like the other examined marine mammals, L. carcinophagus and H. leptonyx had platycoelous centra. These were cylindrical and wider than long or high in the cervical region, where the spinous processes are reduced (Fig. 1d). The thoracic centra have transversely broad articular surfaces, and the rib attachments further limit flexibility (as in cetaceans; Buchholtz & Schur, Reference Buchholtz and Schur2004). The thoracic spinous processes are short (Fig. 2d), but massive transverse processes on the lumbar vertebrae (Fig. 3d) support muscle blocks effecting lateral movements of the hind body, in addition to extensors of the vertebral column (Berta & Sumich, Reference Berta and Sumich1999). Loose inter-lumbar articulations enhance flexibility (Harrison & King, Reference Harrison and King1965) and the lumbar–sacral zygapophyseal contacts are oriented to permit elevation of the sacral complex above the vertebral axis (Howell, Reference Howell1929). The sacrum itself is fully fused (Fig. 3d). The caudal vertebrae are cylindrical with low neural spines (Fig. 4d).
4.b. Pectoral girdle and forelimb
4.b.1. Pectoral girdle
4.b.1.a. Platypterygius
As for all other ichthyosaurs (see McGowan & Motani, Reference McGowan, Motani and Sues2003), the pectoral girdle of P. australis consists of discrete scapulae, coracoids, clavicles and interclavicle. However, only the scapulae and coracoids are documented in this taxon. In our sample, partial scapulae occur in QM F2453, QM F18307, QM F12314 and QM F3348, with a complete example in SAM P44323 which is still embedded in matrix. The general form of the scapulae in these individuals is composed of an expanded, fan-shaped proximal end and narrow, strap-like distal end (Fig. 5b). The proximal articulation with the coracoid and the glenoid contribution is rugose and vascularized, suggesting continuation in cartilage. The coracoids in QM F2299, QM F2453, QM F12314, QM F40823 and QM F40822 are approximately circular in dorsoventral view and bear an anterolateral notch (Fig. 5a). The thickened lateral and medial articular surfaces are pitted and rugose, again indicating cartilage at the scapular-coracoid and intercoracoid joints. The coracoids are inclined at c. 15° dorsal to the horizontal plane when articulated (Zammit, Norris & Kear, Reference Zammit2010).
Forelimb and hind limb material. Platypterygius australis: (a) right coracoid QM F40822 in dorsal view; (b) left scapula SAM P44323 in external view; (c) right humerus QM F2573 in dorsal view; (d) manus of QM F10686 in dorsal/ventral view; (e) ischiopubis QM F40821 in internal view; and (f) left hind limb of QM F18906 in dorsal view. Tursiops aduncus: (g) left forelimb in dorsal view. Dugong dugon: (h) left forelimb; and (i) left pelvis in dorsal view. Neophoca cinerea: (j) left forelimb; and (k) left hind limb in dorsal view. Hydrurga leptonyx: (l) left forelimb; and (m) left hind limb in dorsal view. Abbreviations: c – coracoid; F – femur; f – fibula; H – humerus; IP – ischiopubis; m – manus; p – pelvis; r – radius; s – scapula; t – tibia; ta – tarsus; u – ulna. Scale bar is 10 cm across.
4.b.1.b. Caudally propulsive marine mammals
The glenoid facet in T. aduncus and D. dugon is formed only by the scapula. In T. aduncus, the scapula is triangular in outline, broader than long and lacks an obvious scapular spine (Fig. 5g; Table 1). Conversely, the scapula is longer than broad in D. dugon and bears a prominent scapular spine externally.
Length: width ratios of appendicular elements for each taxon. Ratio of Platypterygius scapula uncertain as the specimen is incompletely prepared and overlain by the ribs. Ranges refer to maximum and minimum values obtained across all specimens for each species. Specimens used for ratios listed in Section 3. Measurements were made using dial calipers.
4.b.1.c. Pectorally propulsive marine mammals
Similar to T. aduncus and D. dugon, the scapula of N. cinerea bears the glenoid. In N. cinerea the scapula bears two scapular spines externally, and is substantially broader than long (Table 1) with a greatly enlarged supraspinous fossa (Fig. 5j); this accommodates the supraspinatus muscle for the pectoral limb.
4.b.1.d. Pelvically propulsive marine mammals
The scapular outline of L. carcinophagus and H. leptonyx is almost oval (Fig. 5l). A single spine was observed on the external surface; Harrison & King (Reference Harrison and King1965) noted that the scapular spine in phocids generally tends to be craniad and reduced.
4.b.2. Humerus
4.b.2.a. Platypterygius
Platypterygius australis has a distinctively robust humerus with expanded proximal and distal extremities and a constricted shaft (Fig. 5c). The proximal head is set at a right angle to the distal extremity, and bears both a massive deltopectoral crest and dorsal trochanter. Additional structures for muscle attachment (e.g. tuberosities or crests) are otherwise absent. There are three distal facets for the anterior accessory element, radius and ulna (although four are recorded in the aberrant specimen QM F2573; Zammit, Norris & Kear, Reference Zammit, Norris and Kear2010). Pitted surfaces on the proximal head and distal surface indicate intervening cartilage.
4.b.2.b. Caudally propulsive marine mammals
The proximal head of the humerus in T. aduncus forms a ball-and-socket articulation adjacent to the small medial tubercle. The humeral shaft is constricted, and the distal extremity is anteroposteriorly expanded to accommodate the broad plane of the fin-like distal manus (Fig. 5g). Two articular facets are present on the distal surface for articulation with the radius and ulna.
The humerus of D. dugon has an elongate humeral shaft that tapers towards its distal end (Fig. 5h); prominent lesser (medial) and greater (lateral) tuberosities are present cranial to the humeral head (James, Reference James1974). Distally, the humerus bears hinged joints with the radius and ulna rather than the flat, synarthotic surfaces observed in T. aduncus.
4.b.2.c. Pectorally propulsive marine mammals
The humerus of N. cinerea is short and massive (English, Reference English1977), but relatively more elongate than in T. aduncus and D. dugon (Fig. 5; Table 1). Extensive muscle attachment surfaces are present across the two enlarged tubercles adjacent to the proximal head, as well as the massive deltoid crest. The proximal articular head comprises a ball-and-socket articulation at the glenoid, with hinged distal contacts involving the radius and ulna.
4.b.2.d. Pelvically propulsive marine mammals
The humeri of L. carcinophagus and H. leptonyx are essentially similar to those of N. cinerea (Fig. 5l). However, the shaft is somewhat shorter (Table 1), the deltoid crest is less robust and the greater tuberosity is not elevated above the humeral head (Howell, Reference Howell1929).
4.b.3. Epipodial elements of the forelimb
4.b.3.a. Platypterygius
Three epipodial elements are usually present in the forelimb of P. australis: an anterior accessory, the radius and the ulna. Both the radius and ulna are polygonal, approximately equal in size and generally broader than long. In contrast, the anterior accessory is longer than broad and smaller than the other epipodials. All of the epipodial components are tightly interlocking (though not fused) and were presumably immobile.
4.b.3.b. Caudally propulsive marine mammals
The radius and ulna of T. aduncus are immobile and variably fused with each other and/or the humerus. Conversely, D. dugon has a flexible elbow joint but with the radius and ulna fused at both the proximal and distal ends. The radius and ulna of both taxa are longer than broad (Fig. 5g, h).
4.b.3.c. Pectorally propulsive marine mammals
The radius and ulna of N. cinerea are not fused and retain a functional elbow joint (Fig. 5j). Both bones are also noticeably anteroposteriorly expanded (the ulna proximally and the radius distally).
4.b.3.d. Pelvically propulsive mammals
The epipodials of L. carcinophagus and H. leptonyx (Fig. 5l) are longer than broad and flattened at their distal (radius) or proximal (ulna) ends. A functional elbow joint is present, and there is no evidence of fusion between individual elements.
4.b.4. Manus
4.b.4.a. Platypterygius
Hyperdactyly is extreme in Platypterygius spp. (Maxwell & Kear, Reference Maxwell and Kear2010), with P. australis manifesting at least nine digits (Fig. 5d). Hyperphalangy is also evident with at least 25 phalanges in the longest digit. The polygonal mesopodium and rectangular phalanges are tightly interlocking, forming a broad ‘pavement’ within the five middle digits (Fig. 5d); the individual phalangeal elements are characteristically broader than long. The distal-most phalanges and additional ossifications in the peripheral digits are ovoid or circular and widely spaced (Zammit, Norris & Kear, Reference Zammit, Norris and Kear2010). There is no evidence of phalangeal fusion, although digit bifurcation does occur within the primary axis of the forelimb (Fig. 5d).
4.b.4.b. Caudally propulsive marine mammals
Hyperdactyly is only reported as a digital anomaly in cetaceans (Cooper & Dawson, Reference Cooper and Dawson2009); however, Phocoena sinus is known to show non-anomalous hyperdactyly within discrete populations (Ortega-Ortiz, Villa-Ramírez & Gersenowies, Reference Ortega-Ortiz, Villa-Ramírez and Gersenowies2000). Hyperphalangy is evident only in T. aduncus among our sampled taxa. Carpal fusion is ubiquitous in both T. aduncus and D. dugon but varies between individuals in T. aduncus, whereas the entire carpus is affected in D. dugon (Fig. 5h). The phalanges are longer than broad in both T. aduncus and D. dugon with the distal-most elements becoming subcircular in T. aduncus; the second digit is the longest in T. aduncus as compared to the fourth in D. dugon.
4.b.4.c. Pectorally propulsive marine mammals
The manus of N. cinerea displays neither hyperdactyly nor hyperphalangy, nor is there evidence of carpal fusion (Fig. 5j). All metacarpals and phalanges are markedly flattened and elongate (Table 1). The first digit is the longest, with subsequent digits decreasing in length sequentially (Fig. 5j).
4.b.4.d. Pelvically propulsive marine mammals
Lobodon carcinophagus and H. leptonyx are almost identical to N. cinerea in their overall carpal and phalangeal morphology with flattened, elongate digits that decrease in length antero-posteriorly (Fig. 5l).
4.c. Pelvic girdle and hind limb
4.c.1. Pelvic girdle
4.c.1.a. Platypterygius
The pelvic girdle in ichthyosaurs consists of the ilia, ischia and pubes (McGowan & Motani, Reference McGowan, Motani and Sues2003), with the latter two elements either partially or entirely fused into an ischiopubis complex in advanced thunnosaurians such as P. australis (Zammit, Norris & Kear, Reference Zammit2010). Although the ilium of P. australis has not yet been described, the ischiopubis lacks any trace of the obturator foramen (QM F40821, Fig. 5e). Proximally, the acetabular surface would have articulated with the ilium, which probably did not contact the axial skeleton based on the morphology of related ophthalmosaurian taxa (see Druckenmiller & Maxwell, Reference Druckenmiller and Maxwell2010; Fischer et al. Reference Fischer, Masure, Arkhangelsky and Godefroit2011).
4.c.1.b. Caudally propulsive marine mammals
The pelvis of T. aduncus and D. dugon is reduced to a single elongate bone. Loss of contact between the ilium and sacrum is also manifest in the undifferentiated sacral series (Fig. 3a, b).
4.c.1.c. Pectorally propulsive marine mammals
Neophoca cinerea possesses a fused pelvis (Fig. 5k) comprising the ilium, pubis and ischium. Elongation of the pubis is thought to accommodate distal migration of muscle attachments and increased leverage of the hind limb (Howell, Reference Howell1929).
4.c.1.d. Pelvically propulsive marine mammals
The pelvis of L. carcinophagus and H. leptonyx displays adaptations indicative of its preeminent role in locomotion (Fig. 5m): the ilium is laterally everted such that the iliac wing of the pelvis is deflected outwards (most extreme in phocines; see Harrison & King, Reference Harrison and King1965; King, Reference King1983 in Berta & Sumich, Reference Berta and Sumich1999; figured by Howell, Reference Howell1929); and both the ischium and pubis are long relative to the ilium.
4.c.2. Femur
4.c.2.a. Platypterygius
The femora of P. australis are remarkably similar to the humeri in morphology: the expanded proximal and distal ends are connected by a narrow, waisted shaft (Fig. 5f); there are prominent dorsal and ventral trochanters (other muscle attachment structures are indistinct); and the distal surface bears three articular facets for an anterior accessory element, the tibia and the fibula.
4.c.2.b. Caudally propulsive marine mammals
The femora have been lost in T. aduncus and D. dugon.
4.c.2.c. Pectorally propulsive marine mammals
The femur of N. cinerea is flattened and has expanded proximal and distal ends (Fig. 5k). The femoral shaft is waisted with prominent greater and lesser trochanters. The hip joint forms a ball-and-socket articulation with the acetabulum, and only two distal articular surfaces are present for articulation with the tibia and fibula.
4.c.2.d. Pelvically propulsive marine mammals
The femora of L. carcinophagus and H. leptonyx are closely comparable with N. cinerea, although the lesser trochanter is absent and the distal articular ends of the bones are considerably more robust (Fig. 5m).
4.c.3. Crus and pes
4.c.3.a. Platypterygius
The tibia and fibula of P. australis are subequal in size, and wider than long (Fig. 5f). The epipodial row also contains an anterior accessory element. The individual epipodial and tarsal bones are polygonal, with the distal phalanges being more rectangular; all elements are broader than long. At least five digits were present in the hind limb (although none have yet been found complete), with synarthotic articulations and no evidence of fusion.
4.c.3.b. Caudally propulsive marine mammals
The crus and pes have been lost in T. aduncus and D. dugon.
4.c.3.c. Pectorally propulsive marine mammals
The tibia and fibula of N. cinerea are elongate (Table 1; Fig. 5k) and fused proximally to form an offset articular surface. A functional knee joint is present. The metatarsals and phalanges are elongate and flattened, and the first digit is typically the most robust. Pedal digits are all of a closely comparable length (Fig. 5k).
4.c.3.d. Pelvically propulsive marine mammals
The crus of L. carcinophagus and H. leptonyx possess elongate tibiae and fibulae (Table 1), a functional knee joint and elongate and flattened metatarsals and phalanges (Fig. 5m). The tibia and fibula were fused proximally to form a level surface. Of the five digits in the pes, digit I is the most robust with digits I and V being longer than the remaining digits (Fig. 5m).
5. Discussion
5.a. Comparisons between Platypterygius australis and other ichthyosaurs
Much of the postcranial skeleton of P. australis is consistent with other post-Triassic ichthyosaurs. The shape of the vertebral centrum (see McGowan & Motani, Reference McGowan, Motani and Sues2003; Buchholtz, Reference Buchholtz2001 a) and the lack of pronounced structural regionalization and fusion of the atlas-axis complex are particularly compatible with thunnosaurians (McGowan & Motani, Reference McGowan, Motani and Sues2003). Furthermore, a marked reduction in the overall centrum proportions, especially in the caudal region of P. australis (Zammit, Norris & Kear, Reference Zammit2010), accords with the presence of a fleshy tailfin, although the proportions and shape of this appendage remain unknown.
The shoulder girdle of P. australis consists of four elements as unanimously observed within Ichthyopterygia (McGowan & Motani, Reference McGowan, Motani and Sues2003). Conversely, reduction of the pelvic girdle is extreme with Athabascasaurus (Druckenmiller & Maxwell, Reference Druckenmiller and Maxwell2010) and Sveltonectes (Fischer et al. Reference Fischer, Masure, Arkhangelsky and Godefroit2011) being the only other ichthyosaur taxa that are known to have possessed a fully fused ischiopubis. The forelimb of P. australis, as well as that of other Platypterygius spp., is likewise unique in exhibiting massive dorsal and ventral trochanters, characteristically rectangular phalanges and multiple accessory digits (Kear, Reference Kear2003; Zammit, Norris & Kear, Reference Zammit, Norris and Kear2010). The femora of P. australis are also distinctive in their unique morphology of the dorsal and ventral trochanters (Maxwell, Zammit & Druckenmiller, Reference Maxwell, Zammit and Druckenmiller2012), and number of distal articular facets (Zammit, Norris & Kear, Reference Zammit2010).
5.b. Comparisons between Platypterygius australis and mammalian analogues
The disc-like vertebral centra observed in P. australis (and all post-Triassic ichthyosaurs; Buchholtz, Reference Buchholtz2001 a) are most similar to the compact cervical centra of T. aduncus and D. dugon. In these caudal propulsors, the short, stiff neck is thought to assist in streamlining to reduce movement of the head and to resist the effects of drag (Hildebrand, Reference Hildebrand1974; Alexander, Reference Alexander1975). Fusion of the atlas and axis (see Wade, Reference Wade1990; Maxwell & Kear, Reference Maxwell and Kear2010; Zammit, Norris & Kear, Reference Zammit, Norris and Kear2010) eliminates rotation at the atlantoaxial joint (Osburn, Reference Osburn1903); this is common among cetaceans (Berta & Sumich, Reference Berta and Sumich1999), which usually exhibit fusion of the cervical vertebrae to decrease flexibility (a notable exception is Inia; Best & da Silva, Reference Best and da Silva1993). Conversely, the non-fused cylinder-shaped cervical vertebrae of pinnipeds such as N. cinerea imply flexibility of the head (Buchholtz & Schur, Reference Buchholtz and Schur2004) and might reflect the specific use of the neck for abrupt turning during pectoral-limb-driven swimming (Fish, Reference Fish2004).
All ichthyopterygians lack pronounced structural regionalization throughout the vertebral column (McGowan & Motani, Reference McGowan, Motani and Sues2003), a feature that is conspicuously shared with samples from our caudal propulsor models of T. aduncus and D. dugon. Marked reduction in centrum proportions towards the fluke region in P. australis (Zammit, Norris & Kear, Reference Zammit, Norris and Kear2010) corresponds to the presence of a fleshy caudal fin; the maximum length of the tail is however unknown and, indeed, has not yet been recorded in any Cretaceous ichthyosaur (see Kolb & Sander, Reference Kolb and Sander2009; Maxwell & Kear, Reference Maxwell and Kear2010; Zammit, Norris & Kear, Reference Zammit2010).
The pectoral girdle complex of P. australis is radically different to the equivalent structures in marine mammals, and is therefore of limited utility in functional comparisons. The obvious reduction of the pelvic girdle on the other hand is compatible with tail-driven swimmers such as T. aduncus and D. dugon versus paraxial swimmers such as N. cinerea, L. carcinophagus and H. leptonyx. Critically however, the hind limbs in the seal taxa examined here are also used in locomotion (terrestrial in otariids and aquatic in phocids; English, Reference English1976; Berta & Sumich, Reference Berta and Sumich1999). Terrestrial mobility has been eliminated along with the functional pelvis in cetaceans and sirenians (the pelvic girdle has been entirely lost in the manatee Trichechus inunguis; Husar, Reference Husar1977), possibly to aid in streamlining (Hildebrand, Reference Hildebrand1974) and the development of more effective caudal propulsion (Thewissen et al. Reference Thewissen, Cohn, Stevens, Bajpai, Heyning and Horton2006).
Unlike cetaceans and sirenians, the hind limbs of P. australis were functional but much reduced relative to the pectoral limbs, suggesting a decreased role in locomotion. Strikingly however, the humerus and femur of P. australis are morphologically homogenous (Zammit, Norris & Kear, Reference Zammit, Norris and Kear2010), with characteristically massive dorsal and ventral trochanters (McGowan, Reference McGowan1972) that would have provided extensive insertion areas for muscles (Romer, Reference Romer1956). McGowan (Reference McGowan1972) hypothesized that extreme development of the dorsal and ventral trochanters in Platypterygius spp. was related to enhancing muscle power for forelimb-driven propulsion. However, increased muscle attachment surfaces do not intrinsically imply such a function. Indeed, manoeuvrability (especially at low speeds) and increased joint stability are imperative for swimming in a complex habitat (e.g. Inia; Klima, Oelschläger, & Wünsch, Reference Klima, Oelschläger and Wünsch1980). The distinctly waisted propodial shafts of P. australis (Zammit, Norris & Kear, Reference Zammit2010) result from a proximal expansion of the convex articular head (Fig. 5c, f) and extreme broadening of the distal limb into a wide hydroplane-like structure. In contrast, the humeri of seals such as N. cinerea, which use limbs as hydrofoils, have a prominent deltoid crest that broadens the shaft distally (providing insertion for powerful limb abductors involved in directional changes; Berta & Sumich, Reference Berta and Sumich1999); specific reduction of the dorsal trochanter on the femora of phocids has also been linked with increased flexibility in the hind limb (Berta & Sumich, Reference Berta and Sumich1999).
One of the most distinctive features of the limbs of P. australis and other ichthyopterygians (see McGowan & Motani, Reference McGowan, Motani and Sues2003) is the absence of mobile elbow and knee joints. This parallels cetaceans, where similar synarthotic joints serve to stiffen the limb when acting as a hydroplane (Hildebrand, Reference Hildebrand1974). Some pectorally propulsive aquatic tetrapods (e.g. penguins) also have rigid forelimbs, where they function as structures to generate lift (Fish, Reference Fish2004). The mobile elbow joints of non-cetacean marine mammals are otherwise employed in terrestrial locomotion (otariids; English, Reference English1976), rowing-like swimming (otariids; Feldkamp, Reference Feldkamp1987) or orientation during feeding (sirenians; Cooper et al. Reference Cooper, Dawson, Reidenberg and Berta2007).
The diagnostic rectangular phalanges found in P. australis and some other ophthalmosaurian ichthyosaurs (Fernández, Reference Fernández1997; Bardet & Fernández, Reference Bardet and Fernández2000; McGowan & Motani, Reference McGowan, Motani and Sues2003) have no functional comparison among extant marine mammals, although cetaceans do exhibit broad distal phalanges that might serve to accentuate the volume of water displaced by the limb during movement (Hildebrand, Reference Hildebrand1974). The development of hyperdactyly and hyperphalangy in P. australis could have likewise increased rigidity of the distal limb for improved acceleration performance (Webb, Reference Webb1977) without compromising efficient long-distance swimming (Blake, Reference Blake2004), factors of obvious benefit when manoeuvring in pursuit of prey.
6. Conclusions
Despite repeated arguments for forelimb-driven propulsion in the Cretaceous ichthyosaur Platypterygius (von Huene, Reference von Huene1923; McGowan, Reference McGowan1972; Riess, Reference Riess1986), our examination of the most complete species of this genus, P. australis, revealed only a few structural features in common with the mammalian paraxial swimmers assessed in this study (N. cinerea, L. carcinophagus and H. leptonyx). Indeed, P. australis lacks almost all of the key features correlated with limb-based locomotory systems in aquatic mammals: a flexible cervical region; pronounced morphological regionalization along the vertebral column; well-developed scapula and pelvis; and mobile knee and elbow joints. Conversely, its short stiff neck, rigid distal limbs and compact propodials, as well as the inferred presence of a caudal fin, are all most readily consistent with tail-driven swimming modes, especially those observed in cetaceans which mainly employ the forelimbs for steering and balance. The unusually extensive muscle insertion areas on the propodials of P. australis could therefore reflect a need for powerful forelimb movements to enhance velocity and directional alterations during high-speed swimming. We find no reasonable need to differentiate P. australis or Platypterygius spp. as atypical in their swimming capabilities relative to other post-Triassic ichthyosaurians (contrary to McGowan, Reference McGowan1972). In fact, all advanced ichthyosaurians seem to have been well adapted for a pursuit predator lifestyle, in which the caudal fin would have acted as the main propulsive organ with the limbs imparting manoeuvrability and aiding in acceleration performance.
Acknowledgements
Many thanks to the staff at various museums for arranging access to specimens: R. Jones (AM); P. Stumkat (KKM); S. Hocknull, P. Couper and H. Janetzki (QM); and B. McHenry, M.-A. Binnie, D. Stemmer, C. Kemper, P. Horton and C. Kovach (SAM). We are also indebted to the anonymous reviewers whose comments greatly improved this manuscript. This work was supported by the Mark Mitchell Research Fund (MZ and BPK) and Australian Research Council (BPK).
