Myth of the QANTAS leap: perspectives on the evolution of kangaroo locomotion

Abstract

The distinctive QANTAS ‘flying kangaroo’ motif of Australia’s national airline signifies what many people regard as the pinnacle of kangaroo evolution—a large-bodied marsupial specialized for endurance-hopping. However, while almost all extant macropodoids (the crown group including kangaroos) use hopping gaits to some extent, the fossil record reveals that the locomotory capabilities of extinct macropodoids were comparatively diverse. The earliest recognized Oligocene–middle Miocene macropodoids probably employed quadrupedal bounding, climbing and slower speed hopping as their primary modes of locomotion. Yet, all were apparently small-bodied (<12 kg), with larger-bodied (>20 kg) forms not appearing until the late Miocene coincident with increasing aridity and the spread of openly vegetated habitats. Hopping is functionally problematic at larger body sizes. Consequently, the later radiation of macropodids (kangaroos, wallabies and their relatives) achieved an optimal mass for efficient higher-speed hopping at ∼35 kg, with a theorized extreme limit of ∼140–160 kg. Modern kangaroos otherwise approach the peak mass range for such gaits at ∼50–90 kg, with the gigantic Pliocene–Pleistocene species of Protemnodon (‘giant wallabies’) at ∼100–160 kg likely being predominantly quadrupedal, and sthenurines (short-faced kangaroos) at ∼50–250 kg seemingly using bipedal striding. Here, we review the fossil evidence of macropodoid locomotion over the last ∼25 million years, and present preliminary analyses of limb bone and tarsal metric data. These indicate that the higher-speed endurance-hopping typical of modern large-bodied kangaroos was probably rare or absent in all but a few crown macropodoid lineages. The intrinsic gait variability of macropodoids has therefore diminished with Late Pleistocene megafaunal extinctions. As a result, the famous QANTAS ‘flying kangaroo’ actually depicts only one of what was once many successful locomotory strategies employed by macropodoids to conquer a range of terrestrial and arboreal habitats.

Christine M. Janis [[email protected]], Bristol Palaeobiology Group, School of Earth Sciences, University of Bristol, Bristol BS8 1RJ, UK; Adrian M. O’Driscoll [[email protected]], Centre for Anatomical and Human Studies, Hull York Medical School, University of York, York YO1O 5DD, UK; Benjamin P. Kear [[email protected]], The Museum of Evolution, Uppsala University, Norbyvägen 16, Uppsala SE-75236, Sweden.

Keywords:

• Macropodoidea
• hopping gaits
• Oligocene–Miocene
• Pliocene–Pleistocene
• limb morphology
• body size

KANGAROOS are icons of Australia, as epitomized by the Eastern Grey Kangaroo, Macropus giganteus Shaw, 1790. Macropodidae (sensu Kear & Cooke 2001, Jackson & Groves 2015, Den Boer & Kear 2018, Eldridge et al. 2019, Beck et al. 2022, Westerman et al. 2022)—the crown clade encompassing kangaroos, wallaroos, wallabies (including the swamp wallaby, Wallabia bicolor [Desmarest, 1804], extinct ‘giant wallabies’ Protemnodon Owen, 1874a and Congruus McNamara, 1994, rufous hare-wallabies, nail-tail wallabies, rock-wallabies and forest wallabies), pademelons, tree-kangaroos, the quokka, Setonix brachyurus (Quoy & Gaimard, 1830), the banded hare-wallaby, Lagostrophus fasciatus (Péron & Lesueur, 1807) (Lagostrophinae), and extinct short-faced kangaroos (Sthenurinae)—also constitutes the quintessential ‘eco-model’ for adaptive specialization towards higher speed endurance-hopping (e.g., Dawson & Taylor 1973). Yet, despite this pervasive popular image of the QANTAS ‘flying kangaroo’, the locomotory evolution of kangaroos and their more distant crown clade relatives (Macropodoidea: Fig. 1) is surprisingly diverse with documented fossil and extant species evidencing a range of bipedal, quadrupedal, saltatorial (or ricochetal = ‘bipedal saltation’ sensu Marshall 1974, p. 174), pentapedal, ambulatory, cursorial and scansorial locomotor modes, as well as burrowing behaviours developed over the last ∼25 million years (Ma).

Figure 1. Time calibrated phylogeny of extant and extinct (†) macropodiforms modified from Westerman et al. (2022), Den Boer & Kear (2018), and Prideaux (2004).

Long distance endurance-hopping is only employed by larger-bodied kangaroos (see discussion in Janis et al. 2014). Their functional and physiological parameters are therefore geared towards hopping efficiency, with up to 50% of the expended energy stored in the hind limb extensor tendons and used to power successive leaps (Biewener & Baudinette 1995). Although many large quadrupedal cursorial mammals (e.g., horses) use similar adaptations, and may exhibit comparable levels of extensor tendon energy storage (see Dawson & Webster 2010), their requisite oxygen consumption increases exponentially with speed. Conversely, these oxygen costs plateau in kangaroos, such that increasing speed incurs no added energy expenditure (Dawson & Taylor 1973, Baudinette 1989, 1994, Baudinette et al. 1992). This could reflect the characteristic ability of hopping kangaroos to increase their stride length without modifying stride frequency (see Dawson & Webster 2010, Thornton et al. 2021). Indeed, the economical energetics of hopping were probably integral to the continent-wide radiation of these larger-bodied macropodids, and especially the species of Macropus Shaw, 1790 and Osphranter Gould, 1842, which today dominate a spectrum of habitats from Australia’s western coastal margin across the intracontinental arid-zone to the eastern Australian Alps, and the northern tropics of New Guinea to as far south as the cool-temperate regions of Tasmania (see chapters in Van Dyck & Strahan 2008). Moreover, all extant macropodoids, except the basally divergent musky rat-kangaroo Hypsiprymnodon moschatus Ramsay, 1876 (Johnson & Strahan 1982, Parnaby et al. 2017), employ ‘short-burst’ hopping to some degree, and the skeletal traits necessary for saltation have even been recognized in the geologically oldest macropodoid fossils (e.g., Woodburne 1967, Flannery et al. 1983, Flannery & Rich 1986, Cooke & Kear 1999, Kear et al. 2001a, 2001b, Kear et al. 2007, Kear & Pledge 2008, Black et al. 2014, Den Boer et al. 2019).

While various studies have explored locomotion in larger-bodied kangaroos, the parallel constraints affecting smaller-bodied macropodoids are less well understood. Certainly, potoroos and bettongs (Potoroidae) have insufficient body mass (<4 kg: Seebeck & Rose 1989) to decouple their energy costs from speed (Bennett 2000; although see Dawson & Webster 2010 for an alternative assessment), and extant macropodids of differing body sizes tend to use a range of less energy-efficient slower-speed gaits (Dawson & Webster 2010): (1) quadrupedal progression with the tail tip contacting the ground, or bounding with the tail elevated in smaller-bodied species (Windsor & Dagg 1971); (2) pentapedal progression integrating the muscular tail in larger-bodied species (O’Connor et al. 2014, Dawson et al. 2015); (3) climbing and asynchronous quadrupedal or bipedal walking along branches in tree kangaroos (Windsor & Dagg 1971, Flannery et al. 1996).

In this paper, we review macropodoid locomotion through the lens of the fossil record, and discuss the key skeletal traits indicative of their gait versatility. Our hypothesis is that while the ‘QANTAS leap’ of modern larger-bodied kangaroos provided a unique adaptive advantage, it represents a component of what was (and still is) a manifold gait repertoire underpinning the evolutionary and ecological diversity of the group over an extended period of time.

Institutional abbreviations

AMNH, American Museum of Natural History, New York, USA; FMNH, Field Museum of Natural History, Chicago, USA; NHMUK, The Natural History Museum, London, UK; NMV, Melbourne Museum, Museums Victoria, Melbourne, Australia; NTM, Museum and Art Gallery of the Northern Territory, Alice Springs, Australia; QM, Queensland Museum, Brisbane, Australia; SAMA, South Australian Museum, Adelaide, Australia; UCMP, University of California Museum of Paleontology, Berkeley, USA; UMZC, University of Cambridge Museum of Zoology, Cambridge, UK; WAM, Western Australian Museum, Perth, Australia.

Constraints on kangaroo locomotion

Hopping is an atypical gait mode amongst mammals, but has evolved independently up to eight times: in macropodoids; the late Oligocene–Pliocene South American marsupial clade Argyrolagidae (Abello & Candela 2020); the placental rodent groups Dipodidae, Heteromyidae, Muridae and Pedetidae (McGowan & Collins 2018); and possibly both the Kultarr, Antechinomys lanigar (Gould, 1856 in Gould 1845–1863), a dasyurid marsupial (see Dawson & Webster 2010), and Mesozoic tamirtherians (basally divergent eutherians), such as Zalambdalestes lechei Gregory & Simpson, 1926 (Velazco et al. 2022). With the exception of macropodids and pedetids (spring hares), most of these lineages are characterized by very small body sizes (<500 g). On the other hand, pedetids (e.g., Pedetes capensis at ∼3 kg) can reach body masses equivalent to potoroids (Seebeck & Rose 1989), stem macropodians (Butler et al. 2017; Macropodia = potoroids + macropodids: Westerman et al. 2022), and some diminutive macropodids, such as Lagostrophus fasciatus, which collectively represent the smallest-bodied saltating macropodiforms (Macropodiformes = macropodoids and their stem relatives sensu Den Boer & Kear 2018; see Doronina et al. 2022 for a phylogenetic justification).

The selective constraints on hopping are unknown; however, because both extant hopping rodents and some macropodoids, such as the archetypal Red Kangaroo, Osphranter rufus (Desmarest, 1822), are found in arid regions (Freedman et al. 2020), their locomotory adaptations have been linked to a shared specialization towards xeric settings (e.g., Prideaux & Warburton 2010). Flannery (1982) alternatively suggested that saltation evolved in the earliest macropodoids as a means of traversing densely vegetated environments. McGowan & Collins (2018) further argued that mammalian saltation may have initially developed to avoid predators in closed habitats. Hopping rodents accordingly possess thick extensor tendons from the gastrocnemius, plantaris, and digital flexor muscles of the lower leg that primarily serve to resist forces from leaping to escape predators (Biewener & Blickhan 1988, Biewener & Baudinette 1995). However, larger-bodied macropodids have thin extensor tendons that store elastic energy at the climax of each hop. This capacity is dependent upon body mass, with ∼3 kg representing a minimum functional ‘cut-off’ (Baudinette 1994; although see Christensen et al. 2022 for a counter interpretation based on dipodoid rodents), below which smaller-bodied macropodoids cannot benefit from elastic energy storage and tend to favour quadrupedal progression (Bennett 2000). By contrast, mechanical stresses on the limb bones and tendons increase exponentially with body size, but can be compensated through upright limb postures at body sizes ranging from ∼5–300 kg, and via increased limb diameter with body sizes above ∼300 kg (see Biewener 2005, Dick & Clemente 2017). Upright limb postures optimize the ‘effective mechanical advantage’ (EMA) on joints at the expense of agility (Biewener 2005, Dick & Clemente 2017). Saltating macropodoids must therefore employ a crouching posture at all body sizes to render their EMA independent of mass (Bennett & Taylor 1995); they also cannot reduce the tendon stresses and joint torques experienced by larger quadrupeds employing more upright limb postures. Consequently, extreme mechanical stresses are encountered at much smaller body sizes in macropodoids compared to typical quadrupedal mammals, and while larger-bodied macropodids could compensate through thickening their extensor tendons, this would reduce the ability to store elastic energy (McGowan et al. 2008a).

Bennett & Taylor (1995) estimated that the optimal body mass for saltating macropodids is ∼50–60 kg. Bennett (2000) further suggested that kangaroos experience locomotory energy/power imbalances at body masses exceeding ∼35 kg. Importantly, this brackets the average adult size ranges (∼35–60 kg) of the largest kangaroos, including the Western Grey Kangaroo, Macropus fuliginosus Desmarest, 1817, Macropus giganteus, and O. rufus (although some mature male O. rufus may reach up to ∼90 kg: see Silva & Downing 1995, Freedman et al. 2020). McGowan et al. (2008b) alternatively posited a maximum tendon tensile strength for hopping macropodids at body masses of ∼140 kg; or ∼160 kg based on data from Snelling et al. (2017). Ultimately, therefore, we consider the viability of saltating gaits as having been achieved at the upper body mass limits of modern macropodids (see Thornton et al. 2021 for another recent review), which parallels the optimal body mass of ∼50 kg for high-speed locomotion in extant quadrupedal cursorial mammals (Usherwood & Gladman 2020).

Fossil record of kangaroo locomotion

Late Oligocene to middle Miocene

The highest-speed gait of the earliest macropodiforms was probably a quadrupedal bound, involving near synchronous sequential movements of the hind limbs then forelimbs during progression (see Windsor & Dagg 1971). This is evinced by the most plesiomorphic extant macropodoid, Hypsiprymnodon moschatus, which is principally a terrestrial bounding quadruped, but is also able to climb (Johnson & Strahan 1982). Nevertheless, H. moschatus manifests numerous adaptations for hopping, including a ‘stepped’ calcaneum-cuboid facet that serves to restrict tarsal mobility in derived saltating macropodians (Szalay 1994, Bishop 1997). A similar condition is evident in the geologically oldest macropodiforms, as exemplified by isolated postcranial remains (e.g., a cuboid, NMV P167811, calcaneum, NMV P167814, and metatarsal IV, NMV P172962) from the uppermost Oligocene Tarkarooloo Local Fauna (LF) of the Namba Formation in South Australia (Flannery & Rich 1986). Woodburne et al. (1993–1994) correlated the Tarkarooloo LF with Zone D of the laterally equivalent Etadunna Formation in South Australia, which has also yielded the partial skeleton (SAMA P23626, SAMA P23637, SAMA P23821) of Ngamaroo archeri Kear & Pledge, 2008—currently the earliest-known ‘macropodian-grade’ macropodoid represented by articulated hind limb elements. In addition to a ‘stepped’ calcaneum-cuboid facet, N. archeri possesses distinctive limb bone character states that implicate at least some propensity for hopping.

1. Hind limb proportions. Kear et al. (2008, p. 30) reported tibia/femur and metatarsal IV/femur length ratios in N. archeri (= ‘Purtia sp.’: sensu Kear et al. 2008) that were ‘well within the range of modern bipedal saltators’. Furthermore, Kear et al. (2008) concluded that while increasing tibia length relative to that of the femur appears to have scaled with body size, elongation of metatarsal IV was unequivocally present in even the most basally divergent macropodoids (e.g., NMV P172962 from the Namba Formation: Flannery & Rich 1986, p. 438, fig. 12.8–9).

2. Hind limb skeletal morphology. Kear & Pledge (2008) reported that the femur of N. archeri had a domed head oriented perpendicular to the shaft, as in saltating macropodians (Kear et al. 2007). The tibia shaft was also straight (similar to larger-bodied macropodids: Murray 1995, Wells & Tedford 1995, Kear et al. 2001a), and the tibia-fibula diaphyseal contact was elongate, thereby stiffening the lower leg (Flannery & Szalay 1982) for efficient hopping (Kear et al. 2001a).

Although no stratigraphically older macropodiform skeletons have yet been documented, complimentary molecular and total evidence phylogenetic dating analyses imply a nascent divergence of macropodians from other macropodoids at around the Eocene–Oligocene boundary (e.g., Beck et al. 2022, Westerman et al. 2022). This suggests a continuous acquisition of saltating traits throughout the macropodoid crown lineage since the late Palaeogene (sensu Burk et al. 1998, Burk & Springer 2000). Flannery (1989a) alternatively advocated multiple independent appearances of hopping during the Neogene, a hypothesis based on the now outdated classification of H. moschatus as a member of Potoroidae (see Westerman et al. 2022 for more recent clade definitions). Nonetheless, the topological placement of H. moschatus amongst more stem-ward macropodiforms, such as Palaeopotorous priscus Flannery & Rich, 1986 (see Den Boer & Kear 2018), and the potential stem macropodoid clades Propleopinae (giant rat-kangaroos or ‘carnivorous kangaroos’: Eldridge et al. 2019) and Balbaridae (fanged kangaroos: Eldridge et al. 2019) remains ambiguous (e.g., Bates et al. 2014, Butler et al. 2018, Den Boer & Kear 2018), and these groups presumably utilized a range of primary gaits and lifestyles. For example, the putatively omnivorous-carnivorous (Archer & Flannery 1985) propleopines are assumed to have been quadrupedal, based on interpretation of two isolated humeri (SAMA P18846, SAMA P35648) attributed to Propleopus oscillans (De Vis, 1888) from the Late Pleistocene Henschke’s Quarry Fossil Cave in the UNESCO Naracoorte Caves Nation Park of South Australia (Pledge 1981, Ride et al. 1997). These bones display straight shafts with reduced pectoral ridges and deltoid crests, which are similar to the humeri of H. moschatus, prompting Ride et al. (1997) to propose cursorial abilities and reconstruct P. oscillans as an ecologically unique macropodiform pursuit predator. While no equivalent postcranial elements have yet been identified for the late Oligocene–early late Miocene propleopine genus Ekaltadeta Archer & Flannery, 1985 (Wroe 1996), the possible close affinity of this taxon with balbarids (Wroe et al. 1998) could reflect a dichotomy of increasingly bipedal macropodians versus mainly quadrupedal basal macropodiforms (as initially implied by Cooke & Kear 1999, Kear & Cooke 2001), with H. moschatus constituting the sole surviving relic of their ‘primitive’ locomotory condition (sensu Burk et al. 1998).

Unlike propleopines, balbarids are represented by both isolated postcranial elements (e.g., Flannery et al. 1983, Den Boer et al. 2019), and at least five incomplete skeletons referred to the late Oligocene to early Miocene taxon Ganawamaya gillespieae (Kear, Cooke, Archer & Flannery, 2007) (QM F35432; originally attributed to the genus Nambaroo Flannery & Rich, 1986: Kear et al. 2007, Butler et al. 2018), and the middle Miocene Balbaroo nalima Black, Travouillon, Den Boer, Kear, Cooke & Archer, 2014 (including the composite specimens QM F41234, QM F41270, QM F50468, QM F52809). These fossils were recovered from Faunal Zones (FZs) A–C in the Riversleigh World Heritage Area (RWHA) of northwestern Queensland (see Arena et al. 2016). The associated pelvis and hind limb bones of G. gillespieae and B. nalima reveal character state mosaics consistent with extant hopping macropodians—e.g., laterally flared ilia with posteriorly positioned sacroiliac contacts, prominent pubic pectineal processes, domed femoral heads oriented perpendicular to the limb bone shafts, and ‘stepped’ calcaneum-cuboid facets (Kear et al. 2007, Black et al. 2014). However, the conspicuously short and broad pes of G. gillespieae has a well-developed pedal digit I—a trait exclusively shared with H. moschatus (Kear et al. 2007). Furthermore, the straight humeral shaft and dorsoventrally expanded olecranon processes on the massive ulnae (a feature evident in B. nalima: Black et al. 2014) led Kear et al. (2007) to suggest that balbarids relied upon their forelimbs for locomotion.

Jones et al. (2022) supported this hypothesis of quadrupedality with geometric morphometric comparisons of distal humeral shape that clustered G. gillespieae, H. moschatus, and some species of Protemnodon amongst terrestrial quadrupedal marsupials. Interestingly, Kear et al. (2008, p. 29) also reported compatible tibia/femur proportions in G. gillespieae and the ‘smaller-bodied’ (at ∼45 kg: Helgen et al. 2006) Protemnodon species, Protemnodon hopei Flannery, 1992 and Protemnodon tumbuna Menzies & Ballard, 1994, which potentially favoured quadrupedal bounding as opposed to ‘full[y] bipedal’ gaits inferred for the larger-bodied (>100 kg: Helgen et al. 2006) Protemnodon anak Owen, 1874a. Black et al. (2014, p. 25) likewise advocated locomotory diversity in balbarids, with the femoral shaft of B. nalima exhibiting prominent attachments for the quadratus femoris and adductor musculature (incipient in G. gillespieae: Kear et al. 2007) equivalent to the ‘boss[-like]’ muscle insertions found in saltating macropodids. On the other hand, the tibiae of both G. gillespieae and B. nalima had sigmoidal shafts with proximodistally shortened tibia-fibula diaphyseal contacts reminiscent of H. moschatus and tree-kangaroos (Dendrolagini), such as the extant species of Dendrolagus Müller, 1840 (see Groves 1982) and extinct Bohra Flannery & Szalay, 1982 (Kear et al. 2007, Black et al. 2014).

The plesiomorphic short lateral trochlear crest and long condyle-like navicular facet on the astragalus of G. gillespieae and other early Miocene balbarids (e.g., an isolated astragalus [QM F59022], distal section of fibula [QM F59024] and downcurved pedal ungual [QM F59025]) additionally denote transverse flexibility within the foot, and perhaps an ability to climb (Kear et al. 2007, Den Boer et al. 2019). Both G. gillespieae and B. nalima display dorsoventrally compressed caudal vertebrae indicative of prehensile tails (Kear et al. 2007, Black et al. 2014). However, their use of arboreal habitats probably differed, as suggested by linear and geometric morphometric analyses of their tarsal and pedal elements (Janis et al. 2016, Den Boer et al. 2019). These reveal disparate locomotor correlations between balbarids and H. moschatus, species of Dendrolagus, and small-to-medium-sized macropodids, such as Setonix brachyurus (a facultative quadruped: Windsor & Dagg 1971), and the species of Thylogale Gray, 1837 (pademelons), which use hopping to move along trails through dense vegetation: Windsor & Dagg 1971). The largest-bodied balbarid recognized to date, B. nalima (up to ∼12 kg: Butler et al. 2017), alternatively converged on kangaroo tarsal proportions (e.g., the species of Macropus: Janis et al. 2016), suggesting a capacity for more efficient saltating locomotion.

In summary, despite their scant postcranial fossil record, late Oligocene and early to middle Miocene macropodoids clearly employed a gamut of locomotory modes. These ranged from quadrupedal bounding and scansorial progression in hypsiprymnodontines (Hypsiprymnodontinae, musky rat-kangaroos: Westerman et al. 2022), balbarids and propleopines, to hopping gaits (at least at higher speeds) in stem macropodians and potentially some balbarids (Kear et al. 2007, Janis et al. 2016, Den Boer et al. 2019). Kear et al. (2001a, 2001b) and Kear (2002, p. 314) also inferred quadrupedalism (and even ‘limited climbing ability’) in stem macropodids based on postcranial remains of an osteologically immature Ganguroo robustiter Cooke, Travouillon, Archer & Hand, 2015 (QM F30845; originally designated as Ganguroo bilamina Cooke, 1997 by Kear et al. 2001b), and a second indeterminate taxon (QM F50419) from middle Miocene FZ C deposits of the RWHA (see Kear 2002, p. 309, fig. 5A, B); this exhibited short and stocky metatarsals like Dendrolagus and Protemnodon (e.g., P. tumbuna: see Menzies & Ballard 1994, p. 130, fig. 8f). Notably, though, no macropodoid appears to have exceeded ∼12 kg prior to the middle Miocene (Butler et al. 2017). The limb morphologies necessary for specialized endurance hopping were therefore a feature of geologically later larger-bodied taxa, and only appeared with increasing size in sthenurines (Prideaux 2004) and macropodines (Macropodinae = macropodids excluding lagostrophines and sthenurines: Westerman et al. 2022). These clades diversified after the middle Miocene in conjunction with the transition from bunolophodont/low-crowned bilophodont molars of balbarids and other basally divergent macropodoids (denoting ‘potoroine-like’ omnivorous, frugivorous, fungivorous and folivorous diets: Kear 2003, p. 301, Butler et al. 2017), to higher-crowned bilophodont molars in sthenurines and macropodines, which propagated during the spread of xeric habitats in the late Miocene and Pliocene (Couzens & Prideaux 2018, Westerman et al. 2022).

Late Miocene to Pliocene–Pleistocene

The upper Miocene and Pliocene–Pleistocene record of macropodoid postcranial remains is largely limited to sthenurine and macropodine taxa. Potoroids, which diverged from macropodids around the early Miocene (Westerman et al. 2022), are diverse today but have a scant fossil record (Kear & Cooke 2001). Their small body masses (<4 kg) and contrasting hind limb proportions (including notably elongate proximal phalanges relative to macropodids: Janis et al. 2022, Jones 2022), also suggest adaption towards different locomotory requirements.

Sthenurines survived up until ∼40 ka (e.g., Prideaux et al. 2010, Gillespie et al. 2012, Hocknull et al. 2020), and included some of the largest-bodied macropodoids, such as Procoptodon goliah (Owen, 1845), which may have reached ∼230 kg (Helgen et al. 2006). Nevertheless, most sthenurines paralleled the upper size range of macropodines at ∼50–150 kg (Wagstaffe et al. 2022). The dentition of sthenurines indicates browsing habits (Couzens & Prideaux 2018), with dolichocephalic (= skull shape equivalent to macropodines: see Prideaux 2004) late Miocene and Pliocene–Pleistocene forms including Hadronomas puckridgi Woodburne, 1967 (Murray 1991) and the shorter-faced species of Sthenurus Owen, 1874a consuming higher-level foliage (Wells & Tedford 1995), while the brachycephalic Metasthenurus newtonae (Prideaux, 2000), species of Procoptodon Owen, 1874b, Simosthenurus Tedford, 1966, and Archaeosimos Prideaux, 2004 (Simosthenurini: sensu Prideaux 2004) all likely fed on tougher plant matter (Prideaux et al. 2009, Mitchell & Wroe 2019). Pliocene–Pleistocene sthenurines are characterized by reduction of pedal digit V to a ‘splint-like’ vestigial metatarsal, thereby imparting functional monodactyly (Wells & Tedford 1995). In addition, Plio-Pleistocene sthenurines possessed elongate acromion processes on the scapulae, and robust forelimbs with elongate manual digits that would have enabled reaching overhead to pull vegetation towards the mouth (Wells & Tedford 1995).

The phylogenetic divergence of sthenurines from macropodines has been variably estimated from ∼38 Ma (Llamas et al. 2015) to ∼19 Ma (Westerman et al. 2022); the latter correlating with the earliest demonstrable sthenurine taxon, Wanburoo hilarus Cooke, 1999 from middle Miocene FZ C and lowermost upper Miocene FZ D deposits of the RWHA (Archer et al. 2007). Butler et al. (2017) calculated a body mass of ∼8 kg for W. hilarus, and although postcranial remains (QM F31456) were initially attributed from FZ D (Kear et al. 2001a), these have been reassigned to a larger-bodied (12–25 kg: Butler et al. 2017; Wagstaffe et al. 2022) sthenurine taxon, Rhizosthenurus flanneryi Kear, 2002, which is diagnostic for the FZ D Encore LF (Myers et al. 2001, Megirian et al. 2010) at ∼11.63–7.25 Ma (Myers et al. 2017) or ∼12 Ma (Megirian et al. 2010, Couzens & Prideaux 2018). Kear (2002) reconstructed an incipiently ‘sthenurine-like’ pes in R. flanneryi, with astragalar trochlear crests oriented parallel to the long-axis of the astragalar neck (denoting a ‘hinge-like’ ankle and shift of the body weight to over digit IV), robust metatarsals, and compact phalanges supporting ‘hoof-like’ pedal unguals. These features are compatible with H. puckridgi (see Murray 1995) from the latest-late Miocene Alcoota LF of the Waite Formation in the central-southern Northern Territory (radiometrically dated at >5.84 Ma, or 7–8 Ma based on faunal correlations: Megirian et al. 2010, Couzens & Prideaux 2018). Significantly, the Alcoota LF has also produced pedal elements (numerous calcanea, astragali and a navicular [UCMP 67082]) attributed to the earliest known macropodine Dorcopsoides fossilis Woodburne, 1967 (e.g., Prideaux & Warburton 2010), which Woodburne (1967, p. 69) interpreted as ‘most close [in] proportion and characteristics to those found in Dorcopsulus vanheuri [(Thomas, 1922)]’. The cuboids, metatarsal IV and other bones also had ‘closest resemblance’ to Setonix brachyurus (Woodburne 1967, p. 73). Woodburne (1967, p. 105) further identified isolated phalanges from the Alcoota LF as an indeterminate ‘small macropodid’ based on their ‘longer and more massive’ proportions relative to D. fossilis.

The body size differentiation amongst Alcoota LF sthenurines and macropodines is stark, with H. puckridgi estimated at ∼73 kg, versus only ∼11 kg for D. fossilis (Wagstaffe et al. 2022). However, their divergence timeframes coincide with macropodines originating at ∼18–19 Ma (Westerman et al. 2022), and the basal split of dorcopsins (Dorcopsini = forest wallabies: Prideaux & Warburton 2010), dendrolagins and macropodins (Macropodini = ‘true kangaroos’ sensu Couzens & Prideaux 2018) occurring at ∼12–14 Ma (Westerman et al. 2022). These groups also use contrasting slower-speed gaits, with quadrupedal progression involving the tail as an arched prop in species of Dorcopsis Schlegel & Müller, 1845, and slow walking or bounding in Dendrolagus and Setonix brachyurus (Windsor & Dagg 1971). Pentapedal propulsion integrating the muscular tail as a ‘limb’ (O’Connor et al. 2014) is variously present in species of Onychogalea Gray, 1841, Wallabia bicolor (Desmarest, 1804), Notamacropus Dawson & Flannery, 1985, Macropus and Osphranter (Macropodini), all of which have elongate tibiae and employ hopping at higher speeds (Dawson et al. 2015).

Pliocene–Pleistocene sthenurines

Sthenurines underwent an adaptive radiation from the Pliocene to Pleistocene (Prideaux 2004), which was characterized by specialization towards asynchronous bipedal striding; a hypothesis predicated on linear metric analyses (Janis et al. 2014) and supported by preserved trackways (Camens & Worthy 2019). Skeletally, sthenurines also acquired the capacity for an upright trunk posture (rather than the ‘macropodine-like’ crouch: Bennett & Taylor 1995), which is possibly related to feeding behaviour rather than locomotion (see Janis et al. 2014), as well as medial transfer of the body weight to over the medial side of digit IV (see Wagstaffe et al. 2022; Harcourt-Smith & Aiello 2004 documented analogous medial weight transfer over digit I in early bipedal hominins).

1. Axial elements. The lumbar vertebrae of sthenurines exhibit elongate metapophyses and reduced transverse processes indicating expansion of the multifidous epaxial versus longissimus dorsi epaxial muscles (Wells & Tedford 1995); this would have restricted dorsoventral flexion of the torso and concomitantly inhibited both hopping and quadrupedal locomotion (Janis et al. 2014). Likewise, the caudal vertebral column was seemingly incapable of hyperflexion beneath the body, thus preventing ‘macropodine-like’ pentapedal progression (Dawson 2015).

2. Pelvis. The pelvis of sthenurines incorporates deflected tuber coxae on the ilia that would have accommodated powerful gluteal musculature for balancing on alternating legs during each stride (Janis et al. 2014). The ischial tuberosities also imply substantial hamstring and adductor muscles, with the short, dorsally tipped ischia suggesting a greater habitual angle between the trunk and femur; the massive epipubics similarly denote substantial hypaxial attachments for elevating and stiffening the trunk (Janis et al. 2014).

3. Hind limb. The femora of sthenurines have large, round heads and broad distal condyles to reinforce the hip and knee joints during walking (e.g., Harcourt-Smith & Aiello 2004). The tibiae are elongate like those of larger-bodied macropodines, but more robust in overall form (Jones 2020).

4. Tarsus. The astragali of late Miocene (e.g., Hadronomas puckridgi: Murray 1995) and Pleistocene sthenurines have deep trochlear grooves and high medial trochlear ridges that would have restricted mobility in the parasagittal plane over digit IV (Janis et al. 2014). The astragalar trochlear groove accommodated a deep plantar process from the distal end of the tibia, which is comparable to the condition in quadrupedal cursorial mammals (Janis et al. 2014). The astragalar fibula facet is expanded, and the calcaneum has an elongate cubonavicular facet to distribute increased body weight (Janis et al. 2014). The calcaneal tuberosity (= calcaneal ‘heel’) is otherwise proximodistally shortened, but distally wider than those of macropodines for insertion of a robust calcaneal tendon (= Achilles tendon) of the gastrocnemius.

5. Pes. The articular surfaces on metatarsal IV and the pedal phalanges of all sthenurines (e.g., Murray 1995, Wells & Tedford 1995, Kear 2002) are transversely broad, and the proximal and intermediate phalanges display a distinctive ‘I-shape’ for enhanced weight-bearing. Their bending resistance was probably compatible to larger-bodied macropodines (Wagstaffe et al. 2022), but the ligament suspensory system around the metatarsal/phalangeal joint was more extensive, presumably bolstering the foot for unilateral weight-bearing (Wagstaffe et al. 2022).

The forelimbs of Pleistocene sthenurines were specialized for reaching and grasping (Wells & Tedford 1995). This is evident from the scapulae, which possessed large infraspinous fossae, reduced supraspinatus fossae, and prominent coracoid and acromion processes (Wells & Tedford 1995). The olecranon process on the ulna was reduced (as in Rhizosthenurus flanneryi and H. puckridgi: Murray 1995, Kear 2002), thereby enhancing elbow extension, but limiting the ability of the forelimb to bear weight (Wells & Tedford 1995). In addition, digits II–IV have elongate phalanges for grasping, while the rounded proximal humeral head, reduced greater tuberosity, enlarged entepicondyle (supporting substantial radial flexors: Wells & Tedford 1995), and a globular distal capitulum and attenuated trochlea imply shoulder and elbow mobility at the expense of stability for weight-bearing (Janis et al. 2020, Jones et al. 2022).

Cumulatively, the evidence for bipedal striding in Pleistocene sthenurines does not preclude saltation in some of the smaller-bodied forms (<100 kg: Janis et al. 2014). Nonetheless, slow-speed bipedal progression was probably favoured relative to quadrupedal or pentapedal locomotion (note that pentapedal locomotion is a derived macropodin trait unlikely to have been present in more basally branching macropodids), and was perhaps the sole gait mode for the largest-bodied (>160 kg) sthenurines (e.g., Procoptodon goliah, whose gigantic size would have exceeded capacity of the leg tendons to sustain hopping: see McGowan et al. 2008b, Snelling et al. 2017).

Pliocene–Pleistocene macropodines

With the exception of dorcopsins, dendrolagins and the basally divergent macropodin Setonix brachyurus (Windsor & Dagg 1971), the specialization of Pliocene–Pleistocene macropodines for hopping coincides with their adaptation towards grazing and grassland environments (Couzens & Prideaux 2018). Indeed, the ecological differentiation of extant grazing macropodins so closely parallels their gait preferences (Janis et al. 2016) that the appearance of endurance-hopping can be regarded as indicative of the clade. However, the species of Petrogale Gray, 1837 (rock-wallabies; Dendrolagini) also became specialized for higher-speed hopping in conjunction with mixed-feeding and complex habitat occupation (see chapters in Van Dyck & Strahan 2008).

The postcranial traits that distinguish larger-bodied macropodins amongst macropodids include: posterior extension of the ischium to enable more powerful retraction of the femur by the hamstring and adductor muscles; a prominent iliopectineal process for insertion of the pectineus (a femoral protractor) and rapid protraction of the hind limb; a dorsoventrally elongate and anteroposteriorly deep tibial crest for attachment of the tibialis anterior, which protracts the lower leg and stabilizes the knee; an elongate and cylindrical calcaneal tuber for increased mechanical advantage during foot retraction by the gastrocnemius; a deeply ‘stepped’ calcaneum-cuboid facet that eliminates lateral movement within the tarsus; and proportional elongation of the tibia, metatarsals and proximal phalanges of digits IV and V to optimize hind limb length (see Janis et al. 2014). All of these features are adaptations towards higher speed saltation and/or endurance-hopping at larger body masses (see Bennett 2000).

By contrast, some atypical Pliocene–Pleistocene macropodins reversed this trend; in particular, the species of Protemnodon (e.g., Janis et al. 2020, Jones et al. 2022), which achieved body masses of ∼45–166 kg (Helgen et al. 2006). The most skeletally well-represented Protemnodon species, Protemnodon anak, is placed as sister to the crown Macropus + Notamacropus + Osphranter clade (e.g., Llamas et al. 2015, Westerman et al. 2022). Yet, its proportionately robust forelimbs and stocky metatarsals IV and V have prompted interpretation as a facultative quadruped (see Kear et al. 2008; Den Boer & Kear 2018, Jones 2020). The short, broad feet of Protemnodon are unique amongst large-bodied terrestrial macropodids, and most closely resemble those of tree-kangaroos (Jones 2020). The correspondingly reduced moment arm of the gastrocnemius muscle for foot retraction would have impeded any application for endurance-hopping. However, trackways attributed to Protemnodon imply some form of saltating gait (Belperio & Fotheringhamm 1990, Carey et al. 2011), which must have approached the biomechanical limits for hopping in Protemnodon species with body masses >100 kg (Helgen et al. 2006, Wagstaffe et al. 2022).

The oldest attributable Protemnodon fossils derive from the lower Pliocene Hamilton LF (4.48–4.62 Ma from Piper et al. 2006; or at a minimum of ∼4.45 Ma from García-Navas et al. 2020) of central-southern Victoria (Flannery et al. 1992, Dawson 2004). Postcranial remains (QM F9075) referred to Protemnodon snewini Bartholomai, 1978 have been described from the mid-Pliocene (∼3.6 Ma: Mackness et al. 2000) Bluff Downs LF of northeastern Queensland, and already exhibit stocky metatarsals (Bartholomai 1978, p. 135, fig. 2) characteristic of the genus. Protemnodon snewini had an elongate tibia like other macropodins (Bartholomai 1978, p. 136, fig. 3), but both the New Guinean Pleistocene species Protemnodon hopei and Protemnodon tumbuna had shortened tibiae comparable to those of tree-kangaroos (Kear et al. 2008).

Figure 2. Scaling of tibia length (proximal to distal articular surface) in Macropodia and other representative macropodoids. Hypsiprymnodon moschatus (Hypsiprymnodontinae), Ganawamaya gillespieae (Balbaridae), the species of Dendrolagus (Dendrolagini) and Macropus ferragus (Macropodini) were excluded from the regression calculations. Unlabelled sthenurine specimens = Sthenurus stirlingi. See Supplementary Data Table S1 for dataset.

Figure 3. Scaling of calcaneum tuberosity length (distal end of the tuberosity to proximal extremity of the sustentacular shelf = area not covered by the astragalus articulation) in Macropodia and other representative macropodoids. Hypsiprymnodon moschatus (Hypsiprymnodontinae), the species of Dendrolagus (Dendrolagini) and Protemnodon (Macropodini) were excluded from the regression calculations. Bold outlines = species using quadrupedal gaits. Unlabelled sthenurine specimens = Sthenurus stirlingi. See Supplementary Data Table S1 for dataset.

Further analyses of the larger-bodied (∼95–160 kg: Helgen et al. 2006, Wagstaffe et al. 2022) Pleistocene P. anak and Protemnodon brehus (Owen, 1874a) suggests that these species may have used their short feet and hooked toe claws to brace against the substrate (Jones 2020). The calcaneum of P. brehus (AMNH 145501) lacks plantar thickening along the tuberosity, which otherwise serves to resist rotational forces during hopping in Macropus giganteus (Wagstaffe et al. 2022). The humeral joint surfaces of P. brehus (AMNH 145501) and P. anak (NMV P39105, NMV P223198, NMV P229318) also indicate greater weight-bearing capacity; the ovoid humeral head and prominent greater tuberosity implying more parasagittally restricted shoulder mobility (Janis et al. 2020), while the squared capitulum and trochlea would have limited elbow rotation (Jones et al. 2022). We therefore interpret Protemnodon sensu lato as having reverted towards facultative quadrupedal locomotion. Their large size and ‘Macropus-like’ elongate tibiae potentially facilitated rearing-up for high-level browsing (Flannery 1982), which accords with their dental morphology (Dawson 2006). Finally, the closely related extinct macropodin Congruus kitcheneri (Flannery, 1989b) from the Late Pleistocene (75–44 ka: Roberts et al. 2010) deposits of Mammoth Cave in southwestern Western Australia, might have been able to climb (see Warburton & Prideaux 2021). Jones et al. (2022) similarly interpreted the distal humeral morphology of Protemnodon otibandus Plane, 1967 (UCMP 70059; a small-bodied Protemnodon species from the latest Pliocene Awe LF of the Otibanda Formation in northern Papua New Guinea) as possibly implicating arboreal behaviour.

Morphometric predictors of kangaroo locomotion

Because the postcranial fossil record of macropodiforms is largely fragmentary, many recent studies have favoured linear or geometric morphometric approaches to analyse proxy elements as predictors of locomotory habits in extinct forms (e.g., Janis et al. 2014, Janis et al. 2016, Den Boer et al. 2019, Janis et al. 2020, Jones et al. 2022). Collectively, these data show that limb bone and tarsal proportions are especially informative. We therefore compiled bivariate plots of log-transformed tibia and calcaneal tuberosity lengths (Supplementary File 1) based on measurements collected by CMJ for the unpublished work of Jones (2020) and O’Driscoll (2020). The tibia and calcaneum constitute the principal weight-bearing elements within the ‘non-serial’ (digit IV–V dominant) hind limb of macropodoids (see Marshall 1974, p. 174), and can be used to illustrate the segregation of hopping traits against increasing body mass over time.

Tibia length

Our visualization of data from Jones (2020) demonstrates that all macropodoids (including stem-group balbarids: see Kear et al. 2008) exhibit proportional elongation of the tibia (Fig. 2). McGowan et al. (2008b) and Doube et al. (2018) reported that tibia length is positively allometric in macropodoids, although Thornton et al. (2021) detected intraspecific negative allometry in Macropus fuliginosus and isometry in Osphranter rufus. We attribute this result to size-skewing (see Hansen & Bartoszek 2012) based on observations of positive allometry (exponent = 0.405 ± 0.028) in extant macropodians (excluding the arboreal species of Dendrolagus), which becomes isometric (exponent = 0.348 ± 0.023; isometry = 0.333) following exclusion of larger-bodied (>35 kg) macropodins (the species of Macropus and Osphranter: Fig. 2). The positive allometry expressed by extant kangaroos is therefore an artefact of extremely elongate tibiae in larger-bodied species, and is not reflective of a general trend within the clade. Furthermore, isometry is maintained (exponent = 0.337 ± 0.025) even when the largest-bodied fossil taxa are included (the species of Protemnodon and sthenurines) but Macropus and Osphranter are excluded (Fig. 2).

The implications of increasing tibia length incorporate elongation of the stride and calcaneal tendon—the primary tendon for elastic energy storage (Biewener & Baudinette 1995, Snelling et al. 2017). Tendon elongation maximizes volume and energy potential (Pollock & Shadwick 1994) without increasing stress, which is a function of tendon cross-sectional area (Biewener & Patek 2018). The species of Macropus and Osphranter are mostly above the ∼35 kg size optimum for effective saltation (Bennett 2000), and thus might have evolved elongate tibiae to maintain tendon energy storage and prevent rupture during high-speed hopping. Interestingly, the basally divergent (see Prideaux 2004) smaller-bodied sthenurines Hadronomas puckridgi and Sthenurus andersoni Marcus, 1962 also possessed tibia lengths that are comparable to equivalent-sized macropodins, and therefore could have retained hopping gaits. By contrast, the similarly-sized Procoptodon gilli (Merrilees, 1965) had a relatively short tibia. Likewise, the gigantic Late Pleistocene kangaroo, Macropus ferragus Owen, 1874a, fits the tibia length regression of typical macropodins, and was thus seemingly adapted for saltation despite its massive body size. However, the relatively shorter tibiae of other sthenurines and large Protemnodon species does not support their use of saltation, at least to the same extent as endurance-hopping macropodins.

Calcaneal tuberosity length

Thornton et al. (2021) found calcaneum length to be positively allometric amongst macropodoids. O’Driscoll (2020) specifically correlated this positive scaling with calcaneal tuberosity length (see also Janis et al. 2021). Our plot likewise demonstrates positive allometry (exponent = 0.393) for extant and extinct macropodians (Fig. 3). However, most sthenurines deviate with significant negative allometry (exponent = 0.151), while Rhizosthenurus flanneryi retains a ‘macropodine-like’ tuberosity length. These results are consistent even after recalculation (exponent = 0.241) to incorporate the nearest extant sister taxon (Llamas et al. 2015, Westerman et al. 2022), Lagostrophus fasciatus, which renders the size range of our sampled taxa comparable to that of other macropodoids.

The crouched hopping posture of macropodins increases tendon stress, and does not allow for EMA adjustment at the limb joints (see Biewener 2005). This can be countered by elongation of the calcaneal tuberosity (Fig. 4), which enlarges the gastrocnemius moment arm and balances the rotational forces (= torque) within the hyper-flexed ankle as the foot lands in stance phase. Amplification of such forces with maximal body size is indicated by thickened cortical bone deposited along the plantar surface of the calcaneal tuberosity, which serves to reinforce the calcaneal tendon insertion in large-bodied macropodins (e.g., Macropus giganteus: Wagstaffe et al. 2022). Conversely, the demonstrably shortened calcanea of sthenurines are primarily composed of trabecular bone with only a thin cortical bone layer along the plantar surface (Wagstaffe et al. 2022). The distal extremity of the tuberosity is also transversely expanded, probably to accommodate a broad calcaneal tendon insertion (see Janis et al. 2014); this is similar to tree-kangaroos, which use powerful ankle flexion for climbing (Warbuton et al. 2012). A broader tendon would be more rupture-resistant, but unfavourable for elastic energy storage (see McGowan et al. 2008a). Notably, transverse expansion of the distal calcaneal tuberosity (albeit to a lesser extent than in sthenurines) is present in the balbarid Ganawamaya gillespieae (see Kear et al. 2007, p. 1157, fig. 10) and some indeterminate middle Miocene macropodids (e.g., Kear 2002, p. 308, fig. 4), indicating that they also possessed relatively thick calcaneal tendons and were therefore not capable of significant elastic energy storage during hopping.

Figure 4. Comparative diagrams of representative macropodiform calcanea in cranial view. A, Macropus giganteus (NMV C5532). B, Dendrolagus lumholtzi (MNH 65258 mirrored from left side). C, Ganawamaya gillespiae (QM F35432). D, Protemnodon anak (NMV P39101.5 mirrored from left side). E, Sthenurus stirlingi (SAMA P22533). Scale bars = 10 mm.

The late Miocene sthenurine R. flanneryi follows the macropodian regression for calcaneal tuberosity length (see Fig. 3), but H. puckridgi approaches the proportions of derived Pleistocene sthenurines, and displays a ‘transitional’ calcaneal bone microstructure (Wagstaffe et al. 2022). Specialization of the macropodoid ankle for endurance-hopping therefore correlates with increasing body size amongst macropodines, with bipedal striding in sthenurines likely preceding reduction of pedal digit V, but being linked with adaptations for unilateral hind limb weight-bearing in late Miocene representatives, such as H. puckridgi (see Janis et al. 2014 for discussion). Curiously, calcaneal tuberosity lengths in the gigantic Late Pleistocene macropodins Protemnodon anak, Protemnodon brehus and Macropus titan Owen, 1938 (sensu Travouillon et al. 2020; the latter with a body mass of ∼175–180 kg: Helgen et al. 2006, Wagstaffe et al. 2022) are proportionally similar to those of smaller-bodied macropodians (Fig. 3), while the calcaneum of at least P. brehus lacks plantar thickening, and that of M. titan is composed almost entirely of cortical bone (Wagstaffe et al. 2022). This implies extreme modification to resist rotational forces affecting the ankle in what was possibly the ‘largest-ever’ hopping kangaroo (body mass up to ∼175 kg, Wagstaffe et al. 2022). Hocknull et al. (2020) recently reported another gigantic Late Pleistocene species of Macropus from South Walker Creek in northern Queensland, which had an estimated body mass of ∼274 kg but this must have been too enormous for effective saltation.

Contrary to larger-bodied macropodins, the demonstrably short calcaneal tuberosity of Pliocene-Pleistocene sthenurines (Fig. 4) was seemingly ill-suited for endurance-hopping at body masses exceeding 100 kg. Hopping necessitates a crouched posture in the locomotor stance phase, and the long calcaneal tuberosity of macropodines acts as a lever arm for the gastrocnemius muscle to offset torque at the ankle joint (see Wagstaffe et al. 2022). The sthenurine calcaneum thus suggests a less-flexed limb posture, with the broad distal tuberosity and correspondingly thick calcaneal tendon imparting inferior elastic energy storage.

Conclusions

While the evolution of macropodiform saltating locomotion apparently originated with the emergence of stem macropodoids, the development of endurance-hopping in larger macropodines was clearly linked to traits that became increasingly advantageous with the spread of openly vegetated environments (Flannery 1982) during the late Miocene (based on divergence estimates from Westerman et al. 2022). Indeed, Marshall (1974, p. 175) recognized that: ‘A kangaroo is an amazingly specialized animal, but its method of travelling by saltation was hardly begun for the purpose of ultimate speed. Rather has it built speed into the locomotory pattern that was already established, probably for some other purpose’. This prescient summation has since been evidenced from the fossil record, which shows that even the stratigraphically oldest macropodoids, including late Oligocene balbarids (e.g., Ganawamaya gillespieae: Kear et al. 2007) and stem macropodians (e.g., Ngamaroo archeri Kear & Pledge 2008, Kear et al. 2008), already possessed skeletal character states consistent with quadrupedal bounding and/or slower-speed saltating progression.

The landmark transition towards endurance-hopping was therefore a unique feature of later Miocene macropodids, and seemingly interlinked with their acquisition of increasingly greater body sizes from the Pliocene to Pleistocene. Once these lineages had exceeded the ∼35 kg optimum body size for hopping (Bennett 2000), there must have been substantial selective pressure to either maximize key skeletal adaptations, such tibia and calcaneal tuberosity length in macropodins, or to adopt alternative gait modes, including quadrupedal progression in the largest-bodied species of Protemnodon, and bipedal striding in sthenurines. Pentapedal locomotion necessitated by tibia elongation (Dawson et al. 2015) in endurance-hopping macropodins is energy expensive (Baudinette 1989, Bennett 2000). Consequently, quadrupedal and bipedal striding gaits might have offered more energy efficient variable speed mobility for the species of Protemnodon and sthenurines, which as higher-level browsers, repurposed the elongate hind legs (that serve to reduce locomotor energy costs at any gait: Strang & Steudal 1990) inherited from their hopping ancestors to both maintain stride length and elevate reach for feeding in the canopy.

The assumption that expanding continent-wide aridity after the Pliocene selectively favoured hopping kangaroos is therefore overly simplistic. Hopping is only one of many gait modes employed by macropodiforms both in the past and today, and should not be regarded as their ecomorphological ‘pinnacle’. What makes modern endurance-hopping kangaroos appear so unusual is the geologically recent extinction of various large-bodied relatives that manifested a wide range of locomotory specializations expressed throughout the group’s long evolutionary history. We are perhaps then in need of a rival Australian airline that covers shorter distances than QANTAS and boasts a novel motif of a striding sthenurine!

Acknowledgments

We thank Robin Beck (University of Salford) for the invitation to participate in this Alcheringa special issue dedicated to Mike Archer. Access to specimens and images was provided by Jin Meng and Judy Galkin (AMNH), Pip Brewer, Roula Pappa, and Nadine Gabriel (NHMUK), Helen Ryan and Kenny Travouillon (WAM), Adam Yates (NTM), Rob Asher and Mathew Lowe (UMZC). Figure graphics were constructed by Melisa Morales-García (www.sciencegraphdesign.com). The reviewers, Guest Editors, and Editorial Board of Alcheringa contributed constructive comments.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplemental data

Supplemental data for this article is available online at https://doi.org/10.1080/03115518.2023.2195895.