Biomechanical energetics of terrestrial locomotion in California sea lions (Zalophus californianus)

The suborder Pinnipedia is composed of three different families of semi-aquatic carnivorous mammals: Otariidae (sea lions and fur seals), Odobenidae (walruses) and Phocidae (true seals). All three families of pinnipeds use their flippers for both aquatic and terrestrial locomotion (English, 1976b; Gordon, 1981; Fish et al., 1988; Deméré and Yonas, 2009). Evolutionary changes to a more aquatic body type have been beneficial for aquatic locomotion, but limiting for terrestrial locomotion (Garrett and Fish, 2015).

The unique structure of otariids indicates adaptations for life both on land and in the water (English, 1976a,b). Being a marine mammal that is both semi-aquatic and semi-terrestrial comes with challenges and compromises. Otariids, like the California sea lion (Zalophus californianus), primarily swim using their foreflippers for propulsion by wing-like flapping, with the hindflippers assuming a passive role (English, 1976b; Friedman and Leftwich, 2014).

California sea lions, including the other otariids, can ‘walk’ on land by alternating their limbs much like a terrestrial quadruped (King, 1964; Hildebrand, 1966; Peterson and Bartholomew, 1967; Tarasoff et al., 1972). The footfalls of a sea lion's walk are evenly spaced in time and are therefore symmetrical (English, 1976b). During a sea lion's walk, the contact of a given hindlimb with the ground is followed by contact of the ipsilateral (lateral) rather than contralateral (diagonal) forefoot, therefore showing a lateral walk sequence (Fig. 1) (English, 1976b). As the sea lion walks, the heels of each of their hindlimbs contacts the ground first, the rest of the foot is dorsally flexed with their digits spread so that their hindlimbs do not contact the forefoot (Hildebrand, 1966). Hildebrand (1966) stated that lateral sequence gaits are ‘superior for avoiding interference between fore and hindfeet’ (Hildebrand, 1966).

Otariids can also locomote by keeping their hindlimbs moving together on one pivot, with the fore flippers moving slightly out of phase for a faster gait (Fig. 1) (Peterson and Bartholomew, 1967; English, 1976b; Beentjes, 1990). Sea lions use their enlarged foreflippers as a propulsive lever when they are moving terrestrially, which may help them conserve energy by avoiding lifting limbs in the vertical direction (Saunders et al., 1953; English, 1976b). In California sea lions, their limbs both support their mass and propel them forward, and the foot pattern is necessary for their forward movement (English, 1976b).

Phocids are more aquatically adapted and therefore are unable to bring their hindlimbs underneath their body to walk. Phocids primarily move forwards by means of spinal flexion and crutching with the forelimbs, but they do not use their hindlimbs for terrestrial locomotion (Garrett and Fish, 2015). This forward progression using spinal undulations in phocids is very different from the quadrupedal gait of otariids (Backhouse, 1961; Kuhn and Frey, 2012; Garrett and Fish, 2015). Phocids demonstrate a lack of speed and have a high energetic cost for terrestrial locomotion (Fish, 2000; Garrett and Fish, 2015; Tennett et al., 2018). This limited performance is a clear example of evolutionary constraints. Morphological adaptations for more aquatic lifestyles although reducing the energetic cost of locomotion in water increases the energy required to move on land (Fish, 2000; Tennett et al., 2018).

Previous studies on the terrestrial locomotion of California sea lions and other otariids have described the kinematics of terrestrial locomotion, (Howell, 1929; 1930; Peterson and Bartholomew, 1967; Beentjes, 1990) but have not examined the energetic cost associated with walking or galloping. Although the energetics of swimming of otariids and phocids has been measured (Williams and Kooyman, 1985; Feldkamp, 1987a,b; Fish et al., 1988), the northern elephant seal (Mirounga angustirostris) is the only pinniped for which the mechanical work of terrestrial locomotion has been calculated (Tennett et al., 2018). The present research addresses the kinematics and locomotor energetics of terrestrial locomotion in California sea lions. The energetics of the California sea lion is compared to the mechanical effort of phocid seals, specifically the harbor seal (Phoca vituline), gray seal (Halichoerus grypus) and the northern elephant seal. It is hypothesized that the quadrupedal gait of the California sea lion will be more efficient, showing a lower mechanical cost of transport (COT), when compared with the undulatory gait of the three phocid seals. Understanding the kinematics and mechanical energetics helps researchers to appreciate the compromises of these amphibious marine mammals and how they locomote on land and in water (Biewener, 1990).

Two female (Cali, Ariel) and one male (Nemo) adult California sea lions (Zalophus californianus Lesson 1828) were examined at SLEWTHS Research Center in Moss Landing, California. The sea lions were trained using classical and operant conditioning and positive reinforcement techniques using food as a reward. The sea lions were maintained on a diet of herring and capelin and were exercised daily and weighed weekly to ensure optimal body condition. All experiments on the sea lions were approved by the West Chester University Institutional Animal Care and Use Committee (Protocol #201601).

The two adult females and adult male were trained to gallop across a level surface while being video recorded. Videos were recorded with a Canon EW-88C camera (24–70 mm lens) at 60 frames s⁻¹. Owing to the size and speed of the animals, the video sampling frequency was adequate to determine the movements of selected points as reported by similar studies (Garrett and Fish, 2015; Tennett et al., 2018). The camera was mounted on a tripod and located approximately two meters away from the sea lion to record galloping strides. The field of view was sufficient to record at least three full strides. Individual sea lions galloped across the mat in the lateral view of the camera. The mat was constructed of 61 cm×61 cm×3.5 cm interlocking foam rubber tiles that were arranged as 3.66 m×1.83 m covering an area of 6.7 m² so that the flipper tips were within the bounds of the mat for video recording and the safety of the animals. Videos where the sea lions did not complete a full stride cycle or walked instead of galloped were not used in the analysis. As opposed to galloping, the walking gait is associated with large lateral head and neck motions (Peterson and Bartholomew, 1967). These lateral motions would add substantial error to energetic calculations that depend on oscillatory motions in the vertical plane [see energetics equation (Eqn 2) below] and could not be compared to the terrestrial locomotion of phocids. Prior to and after the recording sessions a 0.5 m rule was placed in the field of view to scale the video recordings.

Each sea lion was marked with zinc oxide dots on different anatomical points of reference on their bodies. The dots were placed on the center of mass (CM) (Fish et al., 2003), axillary region, iliac crest, ankle, wrist joint and hind flipper tips on both sides of the body. These markings allowed for digital tracking (Fig. 2). CM was approximated from findings in previous studies, as well as through a mass distribution study in the current research. CM was previously determined according to the method of Domining and de Bruffénil (1991), where California sea lions lay on a wooden board resting on a cylindrical pipe. The board was rolled over the pipe until the animal was balanced. The balance point was then measured from the animal's nose to obtain the location of the CM (Fish et al., 2003).

The sea lions were recorded as they galloped across the rubber mat in both directions. Using ImageJ (NIH, v. 1.51 s) and Tracker software (v. 5.1.5), the oscillations of each anatomical point from the sea lions' gallops were tracked and digitized to visualize and measure the path of locomotion (Movie 1). For kinematic analyses, 10 videos for each sea lion were analyzed, digitizing oscillations for all 7 anatomical points. Ten videos with the most clearly defined wave-like oscillations were chosen for each animal. For biomechanical energetic analyses, all 157 videos were used, tracking only CM.

Three species of phocids were used to compare the biomechanical energetics with the California sea lion. These phocids were the northern elephant seal (Mirounga angustirostris Gill 1866), the gray seal (Halichoerus grypus Fabricius 1791), and the harbor seal (Phoca vitulina Linnaeus 1758). Undulation patterns in phocids are the same at various speeds, there is not a distinction between walking and galloping as there is in terrestrial mammals as well as otariids. Undulation of phocid seals was recorded as they moved across a runway towards a fish reward, showing a typical movement pattern that has an incentive. The California sea lions were recorded galloping across a runway with the same fish incentive. Northern elephant seal data were obtained through prior footage of seals locomoting across a sandy beach at Año Nuevo State Park in California (Movie 2). Footage was recorded in 2015 to document the kinematics in northern elephant seal locomotion (Tennett et al., 2018). Ten videos of 10 different male elephant seals were used in the present study. Videos that represent a range of terrestrial locomotion speeds with clearly defined oscillations were chosen. All 10 videos were analyzed with Tracker software to obtain the vertical oscillations of CM. As the elephant seals were wild animals, CM could not be marked. Therefore, the axilla (i.e. posterior insertion of the foreflipper) was tracked as a proxy for CM that could be accurately followed throughout the gallop sequences. When the axilla was not visible because of poor lighting, the eye was followed. The videos were scaled by placing a 0.50 m rule in the field of view of the camera in the location that the seal had just traversed. Body lengths were determined from the scaled videos and measured with ImageJ.

Videos of one captive gray seal and one harbor seal locomoting under trainer control were obtained in 2011 at the Adventure Aquarium in Camden, New Jersey (Garrett and Fish, 2015). One adult gray seal, Kjya, was recorded for 34 locomotor sequences. One adult harbor seal, Spanky, was recorded for 6 locomotor sequences. Videos were tracked and digitized in Tracker software. Videos were tracked following the animals' axilla as a proxy for CM. Videos were scaled using a 0.61 m training pole in the frame.

Seven anatomical points of reference were tracked frame by frame in 10 different locomotor sequences for each adult sea lion. Minimum, maximum, and average velocities (m s⁻¹), along with minimum, maximum and average peak-to-peak (P–P) amplitudes (m) of each marked anatomical point during the gallop cycles. The P–P amplitude was determined by measuring the vertical displacement for each oscillation of the reference points in the sequence and finding the average P–P amplitude for that sequence. The P–P amplitude represents the vertical excursion of each the various tracked points within the sagittal plane of the body. The vertical oscillations of CM with respect to the horizontal movement of the animal is associated with the energy expended (Tennett et al., 2018). All velocity and P–P amplitude data were converted into body lengths (BL) to remove bias due to different body sizes.

Footfalls were studied during the galloping sequence of each individual sea lion to determine the specific gait used. Footfall diagrams were created as the percentage of time each limb is on the ground during one complete gallop stride. One complete cycle started from the time that the left hind flipper contacted the ground to when it contacted the ground a second time (Hildebrand, 1989).

Power output (W) and cost of transport (COT; J kg⁻¹ m⁻¹) are important variables when calculating the biomechanical energetics of a system. The power output is the rate at which work is done during transport of the body at a certain velocity (Tucker, 1970). COT is typically the ratio of metabolic power input to the product of body mass (M) and velocity (Tucker, 1970; 1975; Schmidt-Nielsen, 1972). As metabolic rate for power input was not examined in this study, the power output was used in its place to compute a mechanical COT (Tennett et al., 2018). Power outputs are dependent on the height of the oscillations that a system is making to determine the work expended for that movement. Therefore, P–P amplitude of the oscillations of CM during forward locomotion was used for every individual locomotion sequence by every individual of all four species California sea lion, northern elephant seal, gray seal and harbor seal. Mass-specific power output was also calculated for each species by dividing the animal's mass by their power output (W kg⁻¹).

All statistical analyses were performed in R (v. 4.0.2; https://www.r-project.org/), and all packages used were a part of R. Values for each individual's trials were expressed as mean±s.d. The small sample size reflects the availability of trained sea lions that could be positioned appropriately for detailed data collection. Owing to the low number of animals available, multiple experimental trials were made to provide an adequate statistical sample. For both kinematic and biomechanical analyses, analysis of variance tests (ANOVAs) followed by Tukey HSD tests were done to compare variables. In kinematics, the amplitude, maximum velocity and average velocity were compared against each anatomical body point (Tables S1,S2). In biomechanics, the COT was compared against power output, velocity and amplitude for each of the three sea lions, as well as between each of the four pinniped species (Tables S4,S5). Means±s.d were recorded as well as 95% confidence intervals (DiStefano, 2004). Additionally, least-square regressions and correlation coefficients were used to investigate frequency, as well as relationships between P–P amplitude and both the power output and COT. For biomechanical analyses, the sample size differed for each of the four species. Therefore, regressions were made following each of the species separately to normalize the sample size. In all statistical tests, a value of P<0.05 was considered significant (Whitlock and Schluter, 2015).

A total of 157 videos were recorded for the three adult sea lions. Each of the sea lions that were examined displayed a minimum of two full gallop cycles for each experimental sequence. Overall, mean velocities between the seven anatomical points of reference for all three sea lions together were statistically similar aside from hindflipper tips (d.f.=6, F=7.348, P<0.05). However, the hindflipper tips traveled the fastest during galloping bouts, only showing similar velocities to the foreflipper and the rostrum. The hindflipper tips had a mean velocity of 2.37±0.35 m s⁻¹ for forward movement. The hindflipper tips had the fastest maximum velocity of 11.63 m s⁻¹ during forward galloping (d.f.=6, F=196.2, P<0.05). This was almost 6 m s⁻¹ faster than the next highest maximum velocity displayed by the foreflipper. The foreflipper and the ankle show statistically similar maximum velocities (P<0.05). The anatomical points on both the fore- and hindflippers have the fastest maximum velocities and the slowest minimum velocities for moving the limbs forwards (Tables S1–S3). Minimum velocities for the fore- and hindflipper points were 0 m s⁻¹ considering they rest on the ground at points during the gallop cycle and thus were not in constant motion. The rostrum, CM, axillary region and iliac crest were in constant motion throughout the gallop cycle. CM and the rostrum displayed similar velocity trends over the gallop sequence.

The footfalls of each sea lion showed a transverse gallop, with the right hindlimb touching the ground before the left hindlimb does and the forelimbs also moving slightly out of phase of one another (Fig. 3). The left and right hindlimbs were on the ground for ∼50% of the cycle, while the two forelimbs are on the ground each about 30% of the cycle. There are at least two limbs on the ground at the same time for ∼90% of the cycle (Fig. 3).

All tracked anatomical points oscillated vertically and all points showed a similar oscillation trend throughout the gallop cycle (Movie 1). There were statistical differences in amplitude between the body points (d.f.=6, F=24.64, P<0.05). The greatest maximum P–P amplitude was displayed by the rostrum at 0.36 m (P<0.05). The ankle displayed the smallest amplitude (P<0.05) with a mean P–P amplitude of 0.07 m, which is 68% lower than the mean value of the rostrum. The foreflipper, CM, axillary region, iliac crest and hind flipper tip showed statistically similar mean P–P amplitude values (P>0.05). The hindflipper tip showed the second highest maximum amplitude value after the rostrum (Tables S2,S3).

The power output, mechanical COT, P–P amplitude and velocity were calculated for the three adult sea lions based on the vertical displacements of CM. The pooled average power output for all sea lions was 112.04±18.94 W (Table 1, Table S7).

Nemo's COT was 51% lower than COT in the two females, yet statistic difference between sexes was not recorded as there was only one male (Table 1, Table S6). The two females had similar COT values. The average velocity was statistically different for each sea lion (d.f.=2, F=13.17, P<0.05) (Table 1, Table S8). Amplitude differed for the sea lions individually (d.f.=2, F=9.021, P<0.05). Ariel showed the greatest P–P amplitude for her CM (P<0.05), which was an 18% increase compared to Cali and Nemo. Cali and Nemo had statistically similar P–P amplitudes (P=0.37) (Table 1, Table S9).

The power output and the COT increased for each individual sea lion with vertical displacement about CM. P–P amplitude was highly correlated with COT and power output for each individual sea lion (Table S4).

When corrected for body size, California sea lions showed significantly faster velocities, lower vertical displacements and lower COT (P<0.01) compared with all three phocid species. Power outputs for the three phocid seals were significantly higher than that of the California sea lion (d.f.=3, F=81.29, P<0.05). The northern elephant seal shows the highest power output (P<0.05) (Fig. 4A, Table S5). Average power output for the California sea lions was 96% lower than that of the northern elephant seals and 48% lower when corrected for body size (Table 2, Table S10). The mass-specific power outputs of the gray and harbor seal were significantly higher than the California sea lion as well (P<0.05). Power output for the gray seal and the harbor seal were 46% and 30% higher than the California sea lions, respectively (P<0.05) (Table 2). COT for the three phocid species was significantly higher than COT for California sea lions (d.f.=3, F=40.32, P<0.05) (Table 2, Table S11; Fig. 4B). COT for the three phocids species collectively was 69% higher than COT in the California sea lions. The phocid species had statistically similar COT (P>0.05) (Table 2).

All three phocid seals showed significantly higher P–P amplitudes compared with sea lions when corrected for body size (d.f.=3, F=38.28, P<0.0001) (Fig. 5A, Table 2). The P–P amplitudes of the northern elephant seals, gray seal and harbor seal were significantly greater (P<0.05) than the P–P amplitude of CM of the California sea lions (Table 2, Table S12). Similarly, the length-specific P–P amplitude of the California sea lion was statistically lower (P<0.05) than the values for the phocids. The harbor seal showed P–P amplitudes greater than the gray seal when corrected for body length (P<0.01), but similar to the northern elephant seal (P=0.87). The harbor seal and northern elephant seal had an average length-specific P–P amplitude that was 46% higher than the sea lions. The gray seal had a P–P amplitude 25% higher than the sea lions and 27% lower than the other two seals (Table 2). As P–P amplitude increases, the power output as well as the COT increased with a positive correlation for the California sea lion, harbor seal and gray seal. The northern elephant seal showed no correlation between its P–P amplitude and power output (R²=0.002), as well as P–P amplitude and COT (R²=0.261) (Fig. 5A). Velocities were corrected for body size and showed a significant difference between the phocids and the California sea lions (d.f.=3, F=75.63, P<0.05). The California sea lions had significantly faster velocities than each of the phocid species (P<0.05). There was no correlation between velocity and COT for California sea lions. However, there was a correlation for the harbor seal and the northern elephant seals (Fig. 5B).

A gait is a cyclic motion when moving on land. Gaits are described as either symmetrical or asymmetrical. Symmetrical gaits include various walks and trots, whereas asymmetrical gaits are gallops and bounds (Howell, 1944; Hildebrand, 1966). California sea lions were described previously to gallop with a half-bound gait according to English (1976b). The half-bound is a primitive gait where the hindlimbs move in unison and the forelimbs have a staggered contact with the ground (Hildebrand, 1977, 1989). Observations by Beentjes (1990) noted that Hooker's sea lions (Phocarctos hookeri) gallop with all four flippers placed on the ground and then lifted independently, whereas the California sea lions did not display this pattern. California sea lions can gallop with their hindlimbs moving together in unison, the heels are used as a pivot while the foreflippers advance slightly out of phase (Peterson and Bartholomew, 1967). The head moves within the sagittal plane when galloping, bobbing vertically as opposed to side to side. The sea lions in the present study primarily displayed a transverse gallop where the hindlimbs land on the ground slightly out of phase of each other.

Differences in gait are likely to be due to differing habitats and substrates. California sea lions will use more bounding gaits when moving over rocky terrain (English, 1976b; Beentjes, 1990). As a California sea lion walks, their hindlimbs make short strides as they cannot move independently for a long distance. With each step, the sea lion's hips and shoulders rotate while their head moves sinuously from side to side (Peterson and Bartholomew, 1967). On wet ground, California sea lions will often move forwards using only their forelimbs with their abdomens on the ground and their hindlimbs dragging behind. This movement allows faster strides across wet sand or shallow water (Peterson and Bartholomew, 1967). On smooth ground, a bull California sea lion can ou'trun a man, maintaining a speed of 6.7 m s⁻¹ for a short distance (Peterson and Bartholomew, 1967). At slow speeds, otariids can maintain a terrestrial gallop for long distances. Fur seals have been documented galloping for up to three quarters of a mile (1.2 km) (Bartholomew and Wilke, 1956).

Locomotion by terrestrial carnivores (fissipeds; e.g. dogs, cats) mirrors the same sequences as sea lions for forelimb movements (English, 1976a,b). The foreflippers of sea lions are enlarged when compared with a terrestrial carnivore's forelimb in relation to respective body sizes. Except when dragging the posterior body across wet sand, sea lions extensively move their axial skeleton during terrestrial locomotion because of their aquatically adapted body shape. Fissiped carnivores get all the thrust they need from their hindlimbs when walking (Hildebrand, 1959). Carnivores such as the cheetah and dogs gallop utilizing substantial flexion and extension of the axial skeleton when moving at high speed (English, 1976a,b). The flexion and extension of the axial skeleton of galloping sea lions is similar to terrestrial carnivores. The body design and use of the axial skeleton in sea lions were found to increase energy consumption compared with terrestrial mammals. However, despite their amphibious body design, sea lions can reach high speeds on land while still maintaining their aquatic efficiency (Hildebrand, 1959; English, 1976b). The ability to travel over land with some proficiency may allow otariids to escape predators.

There is a wide range of footfall patterns, with some quadrupeds bearing more mass on their forelimbs or on their hindlimbs (Hildebrand, 1977). Footfall pattern and body mass distribution may allow larger oscillations of CM than predicted with larger energy exchanges (Griffin et al., 2004). California sea lions have their CM positioned anteriorly on the body, close to the pectoral flippers, thus producing relatively large oscillations of their CM (Fish et al., 2003; English, 1976b). Large oscillations are necessary, because a sea lion's limbs are unable to swing freely owing to the elongation of the digits of the flipper. In this study, the hindflipper tips showed the second highest maximum amplitude value after the rostrum, most likely due to the length of the elongated digits reaching above the hindlimb as it rotates around the ankle joint during galloping bouts.

During quadrupedal gaits in mammals, speed and vertical oscillations about CM change even if the average speed is constant and the path of locomotion linear. These changes are responsible for kinetic and potential energy changes, which compose the mechanical work that an animal's muscles perform to maintain locomotion (Cavagna et al., 1976). Walking gaits rely on the out-of-phase fluctuations in vertical position and forward speed of CM (Cavagna et al., 1977; Griffin et al., 2004). During walking, when CM is at its highest vertical position, potential energy is at its greatest and kinetic energy is at its lowest and vice versa. Many animals reduce the muscular work of walking by exchanging the gravitational potential energy and kinetic energy of CM, like an inverted pendulum (Cavagna et al., 1976; Griffin et al., 2004). During running gaits (e.g. galloping), a spring-mass model applies, where kinetic and potential energies are in phase, similarly to a bouncing ball (Cavagna et al., 1976).

These energy fluctuations mainly come from muscular force generated to support an animal's weight and work associated with movement of CM (Biewener, 2003). Quadrupedal animals use bounding gaits to conserve energy during faster locomotion, whereas the exchange between potential and kinetic energies is used to conserve energy during walking gaits. Gait changes conserve energetic costs as faster gaits reduce vertical oscillations of CM and thus allow large quadrupeds to avoid costly aerial periods (Lee and Harris, 2018).

Past studies examined different modes of terrestrial locomotion in various pinniped species, focusing on physiological and behavioral characteristics, but with only few studies on biomechanics (Backhouse, 1961; O'Gorman, 1963; Ray, 1963; Tarasoff and Fisher, 1970; Tarasoff et al., 1972; Gordon, 1981; Kuhn and Frey, 2012; Garrett and Fish, 2015; Tennett et al., 2018). The mechanical energetics of an animal moving on land are associated with the forces that are applied to the ground to support the animal and push forward CM (Cavagna et al., 1977). The energetics of locomotion by an animal can be performed by measurement of metabolic rate from oxygen consumption, respiratory frequency and heart rate (e.g. Williams, 1999). However, the mechanical energetics for terrestrial locomotion can be determined through examination of the physics of an animal's gait (Fish, 1982; Williams, 1999; Tennett et al., 2018). To measure the mechanical power output of pinnipeds moving on land, the P–P amplitude of the oscillating CM and its velocity can be determined (Tennett et al., 2018). The power outputs calculated by Tennett et al. (2018) for 70 northern elephant seals averaged 3290 W at a velocity of 0.6 BL s⁻¹ (2.16 m s⁻¹) and 5530 W at maximum velocity of 0.71 BL s⁻¹ (2.56 m s⁻¹) with a P–P amplitude of 0.31 m.

In the present study, mechanical COT was calculated using the power output from the energetics equation. While this computational model is accurate, there are limitations. The model assumes the animals are running on level ground, which was the case for all captive sea lions studied. This model tracks CM of an animal to get the average power output. Some other parts of the animals may be oscillating with higher power outputs, but only CM was recorded for the power outputs here. The reason for using CM was that the analysis used the parameter M BL⁻¹, which was treated as a constant. However, M BL⁻¹ of an animal is not constant. When the CM was used, the non-uniformities average out. The model assumed that the oscillatory motion of the CM of the animal moved in a perfect sinusoidal fashion. However, the motion was not perfectly sinusoidal although extremely close.

Biomechanics was assessed in three individual captive sea lions. While this gives novel insight into the mechanical energy costs in terrestrial locomotion of California sea lions, further research is necessary for a mechanical energy cost to represent all otariids. Considering the limitations on the availability of trained otariids, any novel research will impact the range of knowledge on terrestrial locomotion in these mammals. There were also limitations for the performance of the individual animals studied. Nemo was the largest of the three sea lions examined and displayed the lowest COT. However, Nemo would not move at high speed when asked to perform gallops. Cali and Ariel galloped as directed. Although these captive sea lions were able to perform gallops in front of the camera, they were only given a short runway in which to reach a high speed. Wild sea lions have more space to gallop and will often gallop upon disturbance, or to charge at a possible threat. Owing to these differences in the motivation for galloping, wild sea lions would most likely gallop at faster speeds than the captive sea lions.

The northern elephant seals in this study were the only wild phocids examined. The captive gray and harbor seals were following commands to receive a food reward and might have behaved differently than their wild counterparts. The harbor seal in this study was overweight for his age and body length, and did not use his foreflippers to aid in forward locomotion. Harbor seals in the wild were observed to move over land with greater ease, utilizing their foreflippers (Garrett and Fish, 2015). The harbor seal had a P–P amplitude similar to that of the northern elephant seals but different from the gray seal. Therefore, as only this harbor seal was examined, its results might not accurately reflect those of wild harbor seals.

Phocid species showed a higher vertical oscillation and thus larger mechanical energetic costs when compared with the California sea lions. Phocids use a slow ‘inchworm’ type of locomotion to move on land without the use of the hindlimbs (Backhouse, 1961; Ray, 1963; Deméré and Yonas, 2009; Garrett and Fish, 2015; Tennett et al., 2018). The high vertical oscillations of phocids is due to locomoting strictly through crutching of the foreflippers and undulations of their axial skeleton without the use of their hindlimbs (Backhouse, 1961; Kuhn and Frey, 2012; Garrett and Fish, 2015; Tennett et al., 2018).

A direct relationship was seen between the P–P amplitude and power output and COT (R²=0.96). The relationship between P–P amplitude of CM and COT infers that having a low COT is dependent on minimizing the vertical oscillations of CM in a locomotor sequence. Having a high P–P amplitude about CM during terrestrial galloping increases the biomechanical work and energetic expenditure than animals with low vertical displacements (Maynard Smith and Savage, 1956). COT was also compared against velocity for all four species. There was no correlation between velocity and COT for the California sea lions, as there was for the northern elephant seals. Faster speeds could require an increase of energy usage for phocids, which cannot locomote on land using their hindlimbs. Maintaining COT through velocity increases could be further proof that California sea lions are more efficient on land and can locomote like a terrestrial mammal. Both otariids and phocids evolved to be highly aquatically adapted and have energetic limitations when trying to sustain rapid terrestrial locomotion (Garrett and Fish, 2015; Tennett et al., 2018). Sea lions, while having the ability to pull their hindlimbs underneath their bodies to gallop quadrupedally, still have limitations for terrestrial galloping. Having aquatically adapted and enlarged limbs causes otariids to have these large oscillations about their CM in order to lift and protract their limbs (Beentjes, 1990). This limitation causes an increased COT but is not a limitation of terrestrial mammals. In terrestrial mammals, the vertical displacement of CM is minimal, lowering their COT. Although a sea lion's unique gait is necessary for its amphibious lifestyle, their gait is costly when compared with its terrestrial mammal ancestors (Beentjes, 1990).

Terrestrial mammals move with less energetic expense than the pinnipeds in this study by locomoting with a lower vertical displacement of CM. The mechanical COT for Asian elephants (Elephas maximus) was determined to be ∼0.2 J kg⁻¹ m⁻¹ over a range of speeds (Genin et al., 2010). With increasing speed, the elephants decrease the vertical oscillation of the CM from about 3 cm to 1 cm (Genin et al., 2010). Although elephants are larger animals, their vertical oscillations of CM were over 9 cm smaller than for California sea lions and nearly 31 cm smaller than for the northern elephant seal. Humans have a comparatively small mechanical COT of approximately 0.05 J kg⁻¹ m⁻¹ (Lee and Harris, 2018). The mechanical COT for a chipmunk and a large dog (100 kg) were 0.46 J kg⁻¹ m⁻¹ and 0.35 J kg⁻¹ m⁻¹, respectively (Heglund et al., 1982). These values, while only slightly lower than that of a California sea lion, are considerably lower than for phocid seals.

Terrestrial mammals can also reduce energetic cost when galloping through the storage and recovery of elastic energy in tendons and ligaments (Biewener, 2003). However, the aquatic body design of pinnipeds does not show evidence for elastic energy storage (Garrett and Fish, 2015). The tendon makeup in the limbs of terrestrial mammals allows for recovery of mechanical work and lower COT compared with pinnipeds (Biewener, 2003).

California sea lions locomote on land with gaits similar to terrestrial mammals (transverse gallop). The ability to pull the hindlimbs underneath the body and stand upright with both sets of limbs entirely supporting the weight of the body reduces the vertical displacement of CM during galloping. This research shows that a lower vertical oscillation of CM allows for lower power outputs and mechanical COT. When compared with phocid species that lack the ability to tuck their hindlimbs underneath their body, sea lions moved over land with higher velocities, smaller vertical amplitudes and lower COT. Quadrupedal galloping of the California sea lion is more efficient than the undulations of phocid seals. However, the aquatically adapted morphology of the California sea lion does incur a higher energetic cost for movement on land compared with fully terrestrial mammals. Thus, with respect to terrestrial locomotion, the sea lions are intermediate between fully terrestrial mammals and phocids. This intermediate position for sea lions means that compromises are made to function in both aquatic and terrestrial environments (Williams, 1999; Fish, 2000).

We thank the staff and animals at SLEWTHS research center who worked with us to make this research possible. This manuscript references the West Chester University Master's thesis of S.J.K. We also thank the thesis committee, Jennifer Maresh and Michael Rosario, who aided with statistical analysis.