Asymmetric gait in locomotion of Hypsiprymnodon moschatus, the most primitive extant macropodoid marsupial

Introduction

One of the most distinctive modes of locomotion among mammals is bipedal hopping, which has evolved at least once in macropodoid marsupials (kangaroos and relatives), four times in rodents and possibly once in extinct argyrolagid marsupials (McGowan and Collins 2018; Jones et al. 2024). Hopping in macropodoids is particularly remarkable for the high energy economy manifested at fast speeds in some species (Dawson and Taylor 1973; Baudinette et al. 1992; Webster and Dawson 2003), as well as the fact that they are the only hopping mammals known to have evolved wherein adult body masses usually exceed 3 kg (Jones et al. 2024). The origins of this charismatic form of locomotion within marsupials, and why it is rare among Mammalia more broadly, remain important unsolved questions.

In terrestrial locomotion, extant ‘ameridelphian’ marsupials (opossums and relatives) typically use symmetric gaits across most or their entire range of speeds (White 1990; Pridmore 1994; Cartmill et al. 2002; Parchman et al. 2003; Reilly et al. 2009; Biknevicius et al. 2013). That is, touch-down and lift-off events of the left and right feet of a limb pair are evenly spaced in time; examples of such gaits include walking and trotting. In contrast, many extant Australian marsupials incorporate asymmetric gaits into their gait repertoire, particularly at higher speeds (Windsor and Dagg 1971; Dagg 1973; Hildebrand 1977; Goldfinch and Molnar 1978; White 1990; Bennett and Garden 2004; Gaschk et al. 2019). Here, touch-down and lift-off events of the left and right feet of a limb pair are not evenly spaced in time, being temporally ‘gathered’ into one part of the stride cycle; that is, more synchronised with one another, as in bounding or galloping (Hildebrand 1977). Within Macropodoidea, the small, early-diverging potoroids employ a combination of quadrupedal bounding and bipedal hopping at fast speeds (Buchmann and Guiler 1974; Baudinette et al. 1993), which may suggest that adoption of this asymmetric quadrupedal gait was a precursor to the evolution of bipedal hopping in macropodids (Burk et al. 1998; Bennett 2000; McGowan and Collins 2018).

The musky rat-kangaroo, Hypsiprymnodon moschatus, is the sole extant member of the family Hypsiprymnodontidae and the most primitive living macropodoid (Burk et al. 1998; Prideaux and Warburton 2010; Westerman et al. 2022). Understanding locomotion in this species can provide important context for interpreting locomotor evolution within Macropodoidea, including the origins of bipedal hopping. Although locomotion in H. moschatus is frequently mentioned in passing by studies of macropodoid evolution, in reality only a single, brief, qualitative, first-hand description of locomotor behaviour has actually been published (Johnson and Strahan 1982, p. 39). That account implied the use of a bounding-like gait and, importantly, that this is used even at fast speeds, without the adoption of bipedalism (Johnson and Strahan 1982). The present study sought to clarify and expand on these prior observations, by using video recordings to provide the first quantitative assessment of gait patterns in H. moschatus. Through comparison to data for terrestrial locomotion in other quadrupedal marsupials, this offers the necessary first step for evaluating ancestral locomotor behaviour in Macropodoidea.

Materials and methods

Results

Across the entire set of observations made here for level terrestrial locomotion, H. moschatus exclusively used an asymmetric, quadrupedal gait (Fig. 1a). In total, 74 ‘foraging’ bouts and 9 ‘escape’ bouts were able to be sufficiently quantified, and Monte Carlo simulations showed that potential imprecision in footfall event detection does not alter the general patterns recovered (Fig. 1b–d). Considering all six of Hildebrand’s (1977) parameters together, foraging and escape bouts are clearly distinguished in gait space (PERMANOVA: F_1,81 = 31.979, P < 0.001). As expected, periods of limb support were proportionally lower in the ostensibly faster escape bouts (Fig. 1b, d; Mann–Whitney U = 18, P < 0.001); the average number of limbs in contact with the ground across the stride was 1.44 ± 0.45 in escape versus 2.62 ± 0.47 (mean ± s.d.) in foraging. Escape bouts also tended to involve a substantial aerial phase (i.e. when all four limbs are not in contact with the ground), up to 40% of the stride, corresponding to ‘extended suspension’, where the forelimbs spread forwards and the hindlimbs spread backwards. Few foraging bouts contained any element of an aerial phase, which were transient.

Fig. 1.

Asymmetric gait use in Hypsiprymnodon moschatus, and comparison to previous observations for other quadrupedal marsupials. (a) Time lapse of two strides from exemplar bout of ‘foraging’ speed locomotion. (b–d) Plots of standard gait parameters as defined by Hildebrand (1977), showing fore- and hindlimb duty factors (b; β_fore and β_hind respectively), fore- and hindlimb lead times (c; λ_fore and λ_hind respectively), and limb coordination in terms of mid-time lag and hindlimb support (d; δ_mid and t_hind respectively). Original observations plotted as circles, with 50% (dark) and 90% (light) contours of a kernel-estimated probability density function fitted to the results of the Monte Carlo sensitivity simulations. (e–g) Plot of same gait parameters as previously reported for other Australian quadrupedal marsupials, shown as convex hulls circumscribing the full spread of observations. Note that data for Perameles sp. were recorded for a limited number of trials (Hildebrand 1977); so, they are unlikely to span the full range of gaits or speeds employed.

Compared with the asymmetric gaits reported for other quadrupedal marsupials (Fig. 1e–g), H. moschatus uses a notably different gait repertoire, consistently involving very small fore- and hindlimb leads across all speeds (Fig. 1c, f), corresponding to a ‘bound’ or ‘half-bound’. Although the threshold between a bound and a half-bound is subjective (Hildebrand 1977), H. moschatus clearly uses a gait closer to a pure bound, where fore- and hindlimb leads are zero (clustering in the top-right corner of Fig. 1c), in turn resulting in the consistent use of extended suspension at fast speeds (Fig. 1d). Other quadrupedal marsupials tend to use a half-bound or alternatively some form of gallop (see also Dagg 1973; Hildebrand 1977; Baudinette 1994), incorporating a substantial non-zero forelimb lag, and periods of extended or gathered suspension (or both).

Beyond the data reported here, a bounding gait in H. moschatus was observed in numerous other locomotor bouts that were unable to be quantitatively characterised. This included very fast bouts where fewer than two strides were recorded in the video frame of view. Even in body repositioning or reorienting manoeuvres, the forelimb and (especially) hindlimb pairs were almost invariably moved together as a single unit. Independent steps of a single forelimb were occasionally observed during stop–start foraging activity (e.g. reaching forwards toward food); even then, whenever the hindlimbs were moved, they moved as a set (see Supplementary Video S3).

All four limb bone measurements exhibited a moderate influence of phylogenetic signal, according to the K statistic of Blomberg et al. (2003), which varied between 0.2699 and 0.4432 (associated P-values all <0.001). Across the marsupials and rodents measured, the humerus of facultatively bipedal species, in general, tends to be shorter and more gracile relative to the femur than in quadrupeds (Fig. 2, Table 1). This appears to be primarily due to differences in allometric intercept, although in each case allometric slopes are also significantly different (Table 1); indeed, the allometric trajectories for lengths converge at large body size when phylogeny is taken into account (Fig. 2a), suggesting that differences in length may be legitimately associated with locomotor mode only in smaller taxa. The 95% prediction intervals for each group’s pGLS fit is broad enough to encompass numerous datapoints from the other group, especially when phylogeny is taken into account (Fig. 2a, c). The proportions of H. moschatus fall between those of the biped and quadruped fits, and within the 95% prediction intervals of both, but consistently fall closer to the expected mean for quadrupeds (purple lines in Fig. 2).

Fig. 2.

Humerus and femur proportions in marsupials and rodents. (a, b) Bone lengths. (c, d) Minimum diaphyseal circumferences. Generalised least squares regressions and 95% prediction intervals for each group are also presented, both (a, c) considering and (b, d) ignoring the effects of phylogeny. Also reported are regression coefficients for bipeds and quadrupeds (expressed in terms of log₁₀-transformed data). For bone lengths, regressions are also presented for marsupial and rodent bipeds separately to illustrate the difference in allometric slope between the two subgroups (F_2,27 = 6.0428, P = 0.021; statistical significance disappears when phylogeny is accounted for, as expected for mutually exclusive groups).

Discussion

To the extent that the present observations permit, Hypsiprymnodon moschatus uses exclusively asymmetric gaits (essentially bounding) throughout its locomotor repertoire. Moreover, there remains no evidence that this species ever employs a bipedal gait, even in very fast escape manoeuvres (Johnson and Strahan 1982). The present study therefore supports the widely held belief that H. moschatus is the only extant macropodoid that does not engage in bipedal locomotion. Among the small quadrupedal marsupials compared here (Fig. 1), H. moschatus uses a gait approximating a pure bound, contrasting with half-bounds or gallops in the other species (Hildebrand 1977; White 1990; Bennett and Garden 2004). The difference from the terrestrial bandicoots is particularly noteworthy, and may reflect the substantial phylogenetic distance between peramelemorphians and macropodoids (Duchêne et al. 2018; Beck et al. 2022), providing sufficient time for divergent patterns of neuromuscular locomotor coordination to evolve. Such a hypothesis deserves quantitative scrutiny in the future (e.g. Cuff et al. 2019).

In terms of other species of small-bodied macropodoid, insufficient quantitative data currently preclude detailed comparison of gait patterns, and how they vary with speed, between H. moschatus and its relatives, although attempts are being made to address this (Westerveld 2024). In the interim, the following qualitative patterns are worth noting. Quadrupedal bounding is employed at slow speeds in potoroos, but is replaced by the bipedal hop at faster speeds (Buchmann and Guiler 1974; Baudinette et al. 1993). A similar situation has also been reported for a range of macropodids, including the ground-dwelling quokka (Setonix brachyurus), some species of pademelon (genus Thylogale) and rock wallaby (genus Petrogale), and the arboreal tree-kangaroos (genus Dendrolagus) (Windsor and Dagg 1971; Baudinette 1977; Dawson et al. 2015; Westerveld 2024). In macropodids with very long hindlimbs, pentapedal locomotion involving the tail replaces the quadrupedal bound as the preferred slow-speed gait (Dawson et al. 2015). It is also noteworthy that springhares (genus Pedetes), rodents native to southern Africa, also employ a quadrupedal bound in addition to bipedal hopping, on the basis of limited observations by Hildebrand (1977). The increased propensity for asymmetric quadrupedal gaits, especially bounding, in association with bipedal hopping at faster speeds supports the notion that quadrupedal bounding was an evolutionary precursor to bipedal hopping in macropodoids (Burk et al. 1998; Bennett 2000; McGowan and Collins 2018).

If the above hypothesis is correct, it begs the question of why H. moschatus itself does not employ bipedal hopping. This may simply reflect the early phylogenetic divergence of Hypsiprymnodontidae from other macropodoids, perhaps 30 million years ago or more (Duchêne et al. 2018; Cascini et al. 2019; Beck et al. 2022; Westerman et al. 2022), prior to a single origin of hopping behaviour. Indeed, H. moschatus retains several plesiomorphic features suggestive of less specialised locomotor habits, including a mobile pedal digit I, sizeable digits II and III, and a primitive tarsal structure (Johnson and Strahan 1982; Szalay 1994). In addition to less differentiation between fore- and hindlimb lengths, H. moschatus also retains humeral and femoral proportions that are more typical of quadrupeds (Fig. 2). One further aspect worth consideration is that the whole-body centre of mass of H. moschatus may be too far forward of the hips for sustained bipedalism to be biomechanically feasible (see also Bennett and Garden 2004). The combination of a more posterior centre of mass and elongate hindlimbs has been previously linked to the evolution of (obligate) bipedalism in terrestrial archosaurs (Bishop et al. 2020), and the same requirements may also have applied to macropodoids. Irrespective of the reason(s) underlying the lack of bipedalism in H. moschatus, anecdotal observations made during the present study suggest that substantially more weight and propulsion is borne by the hindlimbs than the forelimbs in this species. Although it remains to be experimentally verified, this would nevertheless mirror the condition observed in extant macropodids during slow pentapedal locomotion (O’Connor et al. 2014).

Supplementary material

Supplementary material is available online.