Swimming in the California sea lion: morphometrics, drag and energetics

Within the order Pinnipedia, two distinct swimming styles have evolved. The Phocidae, or true seals, undulate their bodies and rear flippers in a manner analogous to the subcarangiform mode of locomotion in fish (Lighthill, 1969). Sea lions and fur seals of the family Otariidae use their foreflippers, a style which resembles that of penguins and sea turtles (Robinson, 1975). These contrasting methods of propulsion suggest functionally separate approaches to the problems of aquatic foraging.

Previous studies on fish demonstrate that the mechanical performances of caudal and pectoral fin propulsion are quite different (Webb, 1975a, 1984; Blake, 1983a). These differences are often reflected in the foraging patterns and choice of habitat among species (Webb, 1984). An ecological separation based upon functional performance may also occur in pinnipeds.

Few studies have investigated the hydrodynamic characteristics, drag and cost of swimming in aquatic mammals. With the exception of several papers on cetaceans (Lang & Daybell, 1963 ; Lang, 1966) and one on seals (Williams & Kooyman, 1985), most hydrodynamic work has been conducted on models or carcasses (Hertel, 1966; Mordvinov, 1972; Mordvinov & Kurbatov, 1973; Aleyev, 1977). Data on the energetic requirements of locomotion are also scarce and, for seals and sea lions, are only available for a single velocity or over a narrow range of velocities (Kooyman, Kerem, Campbell & Wright, 1973; Costello & Whittow, 1975; Oritsland & Ronald, 1975; Craig & Pasche, 1980; Lavigne, Barchard, Innes & Oritsland, 1982). Only Kruse (1975) and Davis, Williams & Kooyman (1985) have addressed the question of how this cost varies with speed. No one has determined the efficiency of marine mammal propulsion.

The purpose of this study was to examine the hydrodynamics and locomotory energetics of foreflipper propulsion in the California sea lion (Zalophus califomi-anus), a representative otariid. Body shape was quantified and compared with other swimming animals to assess their streamlined design. Drag coefficients were determined and permit an estimate of the power output requirements necessary for swimming. The energetic cost of locomotion was also measured. Coupled with the drag measurements, this provided a means of determining metabolic efficiency and the mechanical efficiency of the foreflippers at generating thrust.

Four sea lions (SL1—SL4) used in these experiments were held in seawater tanks at Scripps Institution of Oceanography and fed a diet of mackerel supplemented with vitamins. Animals were weighed weekly and body mass was maintained at ±3% throughout the study.

Direct measurements were made of body and flipper lengths in all animals. Girth at various locations along the body was determined both by direct measurement and from photographs. Surface areas and volumes were calculated by considering each animal as a series of truncated cones based upon the length and girth measurements. Rear flippers were estimated as triangles. Projected area of the foreflippers was determined by tracing their outline on a digitizing pad. Total surface area was calculated as twice the projected area.

Three juvenile sea lions, two females (SL2, SL3) and one male (SL4) were used in these experiments. SL2 and SL3 both maintained a body mass of 22·5 ± 1·0 kg (±2S.E.) over the study period from August to September. Water temperature (T_w) was 26 ± 0·9°C. SL4 was studied on two separate occasions. During the first, from April to May, T_w averaged 19±0·7°C and the animal weighed 18·0 ±0·5 kg. The second set of experiments was conducted during August. SL4 had increased in mass to 27 kg and T_w had risen to 25 ± 0·5°C.

Rates of oxygen consumption and carbon dioxide production were measured as animals swam against a water current generated inside a variable-speed flow channel, 1·1 m in cross-sectional area and 16m in length. Two 0·6-m diameter propellors generated flow up to a maximum velocity of 1·3m s⁻¹. Water velocity was “calibrated prior to the experiments with an electromagnetic flow probe and correlated with pump speed.

A test section was created by partitioning the flume with two grids, 2·5 m apart. These were constructed of 15 mm × 3 mm steel strips welded together to form 13 cm square openings. Strips were oriented ‘edge on’ to the flow to reduce turbulence. No measurable drop in water velocity was detected at any depth inside the test section.

A Plexiglas dome (1·1 m × 0·6 m × 0·3 m) was set into the forward portion of a plywood frame that covered the test section. Animals could only surface to breathe inside the dome which served as an open-circuit metabolic chamber. Ambient air was pulled through the chamber with a vacuum pump. A dry gas meter connected to the inlet port measured air flow (Fig. 1). Samples of expired gas were continuously removed and dried with Drierite before entering the Applied Electrochemistry (AEI) CO₂ analyser. CO₂ was then removed with Baralyme and the gas redried upon entering the AEI O₂ analyser. An online Apple 11+ computer sampled the voltage output of each analyser and provided 1-min averages of the percentage gas concentration. and were computed using the equations of Depocas & Hart (1957). All values were corrected to STPD.

The system was calibrated daily with dry ambient air and checked for leaks by bleeding 100 % N₂ at a known rate into the air entering the dome. The O₂ fraction in the outlet flow was calculated and compared to the analyser reading (Davis et al.1985). Values agreed to within ±0·01 % prior to the start of each experiment. No difference was detected between the measured and theoretical O₂ fraction at any water velocity.

The CO₂ analyser was calibrated by bleeding a mixture of 10 % CO₂ in N₂ into the total flow. This produced CO₂ levels similar to those of the sea lions. Calibrations were conducted with and without water flowing through the test section. At zero water flow, the theoretical and measured CO₂ values agreed to within ±0·01 %. At higher speeds, however, the measured value was slightly lower, indicating that CO₂ was dissolving into the water. Factors were determined for each velocity and used to correct these readings. Adjusted values agreed with those calculated to within ±0·01 % at all flume speeds.

Practice sessions were undertaken for several weeks prior to the actual measurements. Training was considered complete when reproducible values for were obtained. Animals were fasted for 12–16h before each swimming session. At each velocity increment, 10 min was allowed for the animals to reach steady state. Recordings were then made for 10–20 min of continuous swimming. 10-min averages of and provided single data points at each velocity. Only those measurements where the animal swam steadily were used. Flume speeds were varied randomly to avoid systematic errors, and sea lions rested for several minutes between each velocity increment.

Although sea lions showed an elevated at 1·3 ms⁻¹, this was well below their top swimming speed. To simulate higher swimming velocities, each animal ‘s drag was increased. Two nylon cups (12cm × 12cm × 5 cm) were attached with cyano-acrylic glue to the midline of their ventral surfaces at 40 and 60 % of the body length. No interference was observed between the cups and the foreflippers.

Effective swimming speeds were measured by determining each animal ‘s drag, both with and without drag cups, over the range of experimental velocities. This was accomplished by towing the sea lions underwater behind an electrically powered cart (Williams & Kooyman, 1985). The cart travelled around a circular ‘ring’ tank which had inner and outer diameters of 14·5 and 21 m, respectively. Towing velocities ranged from 0·9 to 3·8 ms⁻¹. Water depth was 3·5 m.

Animals were trained to bite a soft neoprene mouthpiece that fitted inside their mouths and conformed to the jaw profile. The mouthpiece was attached to a 6·4 mm nylon line that was secured to a force transducer mounted on the cart. From the transducer, the towing line passed vertically down through a streamlined hydrofoil strut, 90 cm in length, mounted underneath the cart in an orientation that minimized turbulence. At the bottom of the strut a pulley, encased in a streamlined fibreglass housing, turned the line 90° The depth of the pulley was 1 m. The animal was towed approximately 1 m behind the strut to avoid effects of turbulence.

Before each session, the transducer was calibrated with a hand-held dynamometer previously calibrated against a series of weights. Cart velocity was monitored with a magnetic tachometer attached to one of the outer wheels. Outputs from the transducer and tachometer were simultaneously recorded on a Gould strip chart recorder. During towing, an observer ensured that animals remained motionless with both flippers held against the body in a gliding position. Runs lasted up to 30 s at a constant velocity. Only those which produced a steady trace for a minimum of 5 s and where no movement was observed were analysed.

The resting metabolic rates of SL2 and SL3 were measured in a tank, 1·05 m × 1·5 m in cross-section, 2m in length, and filled to a depth of 1 m with sea water. Water temperature was 26°C. Sea lions fasted for 24 h prior to these measurements. A metabolic dome served as an open-circuit chamber and was measured as described above. Animals were periodically active while in the chamber, but often remained quiet for short (10–30 min) intervals. The lowest 10-min average of was taken to represent the resting metabolic rate.

Morphometric data for the sea lions are presented in Table 1. The aspect ratio, which averaged 7·9, was determined as span of the flippers divided by their chord. Flipper chord was taken to be the width of the flippers at the wrist. Based on volume estimates, body density was computed to be, on average, 1·02× 10³kgm⁻³, nearly equal to seawater (1·025×10³ kgm⁻³ at 15°C; Vogel, 1981).

Drag coefficients, determined from the gliding animals and based on total surface area (Cd_s), averaged 0·0039 ±0·0001 (±2S.E.) for SL1 (N = 13), 0·0042 ± 0·0002 for SL2 (N = 31) and 0·0041 ±0·0004 for SL3 (N = 17). There were no significant differences among the three animals. Reynolds numbers associated with these glides were 2·87×10⁶, 2·03×10⁶ and 2·13×10⁶ for SL1, SL2 and SL3, respectively (Table 2).

The resting metabolic rate of SL2 and SL3 averaged 6·6±0·33 (±2S.E.) and 6·5 ± 0·17 ml O₂ min⁻¹ kg⁻¹, respectively. This rate is 1·4 times that predicted by the Kleiber (1961) relationship of M = 3·4W^0·75 (M is in watts, W is body mass).

To determine whether this metabolic elevation was due to a low T_w, an abbreviated study was undertaken several months later at a T_w of 25°C, when SL4 weighed 27 kg. There was no significant difference between the of SL4 and that of SL2 and SL3 up to 1-3 ms^-1 (Fig. 3).

For SL2 and SL3, the cost of transport (CT), defined as the amount of energy required to move a unit of mass a given distance (Schmidt-Nielsen, 1972; Tucker, 1975), reached a minimum at l·8ms⁻¹, corresponding to a relative speed of 1·4 body lengths s⁻¹ (Ls⁻¹) (Fig. 4). At this velocity, CT averaged 0·12±0·005 (±2 S.E.) ml O₂ kg⁻¹ m⁻¹. For SL4, when T_w averaged 19°C, CT continued to decline as velocity increased and showed no distinct minimum. At the highest speed of 2·6ms⁻¹ (2·1 Ls⁻¹), CT averaged 0·20 ± 0·007 ⁻¹ m⁻¹. At a T_w of 25°C, SL4 showed costs similar to those of SL2 and SL3 at comparable speeds up to 1·3 m s⁻¹ (Fig. 4).

Body shape is perhaps the most important variable affecting an animal ‘s swimming performance. Sea lions swim at speeds where fluid flow around their body is characterized by high Reynolds numbers. The Reynolds number (Re), defined as the product of an animal ‘s length and velocity divided by the fluid ‘s kinematic viscosity, is a measure of the relative importance of inertial to viscous forces in this flow (Vogel, 1981). At higher Re values, inertial forces dominate, and flow in the boundary layer changes from laminar to turbulent. As Re increases, the boundary layer begins to separate from the animal, leading to a turbulent wake (Webb, 1975a; Vogel, 1981). This increases the rate of transfer of momentum to the water, removing kinetic energy and slowing the animal ‘s forward progress. Reducing wake size by delaying boundary layer separation is, therefore, of primary importance in minimizing the rate of energy loss during swimming.

At high Re values, streamlining is the most effective way to delay this separation (Webb, 1975a). The ability of an object to reduce drag depends on two factors, its fineness ratio (FR) (length divided by maximum diameter) and the position of maximum diameter along its body (Webb, 1975a; Aleyev, 1977). Direct length and girth measurements show that sea lions are highly streamlined (Fig. 5). Their FRs averaged 5·5 (Table 1), not substantially different from the optimum of 4·5 which gives minimal drag for maximal body volume (Webb, 1975a). Moreover, their shoulder is at 40% of body length. This position is sufficiently far back to ensure laminar flow over the forward portion of the animal and to delay the point at which separation occurs (Hoerner, 1958; Aleyev, 1977). As a result, disturbance to flow is reduced, a smaller wake is formed, and resistive forces are diminished.

This combination of features has resulted in a morphology that closely resembles a technical body of revolution (Fig. 5). The advantage of this streamlined design is shown by examining the drag coefficient. For objects of similar size travelling at comparable speeds, a lower Cd denotes a reduction in total drag and hence in the power required for steady forward movement (Webb, 1975a). Sea lions have drag coefficients based on wetted surface area (Cd,) of 0·0042–0·0039 at Re values of 2·03–2·87×10⁶ (Table 2). These values are comparable to those found for other large aquatic animals including penguins (Cd, = 0·002–0·0044 at Re = 10⁶) (Clark & Bemis, 1979; Bilo & Nachtigall, 1980), the white-sided dolphin Lagenorhynchus obliquidens (Cd, = 0 ·0034 at Re = 9·1×10⁶) (Lang & Daybell, 1963) and harbor seals (Cd, = 0·004–0·007 at Ac = l·6× 10⁶) (Williams & Kooyman, 1985). In contrast, the Cd₃ values of a streamlined spindle with an FR of 5·5 and a fully turbulent boundary layer were calculated to be 0·0046 and 0·0043 at Re values of 2·0 and 2·9×10⁶, respectively (Hoerner, 1958; Webb, 1975a). The lower values for sea lions demonstrate that they maintain a portion of laminar flow over the forward surface at these speeds.

These calculations describe the drag forces acting on sea lions and reflect the power requirements necessary for overcoming this resistance. When the sea lion is actively stroking, however, total drag will be slightly greater due to resistive forces acting on the foreflippers. These include frictional and pressure drag components as well as interference drag that arises through an interaction between an appendicular propulsor and the body (Blake, 1981). Because foreflippers generate propulsive forces that can be resolved into horizontal thrust and vertical lift components (English, 1976; Feldkamp, 1985), an induced drag component associated with lift formation will also be present (Lighthill, 1975). However, since these forces arise only when the flippers are extended, they are likely to be small in comparison to total body drag (Webb, 1975a; Blake, 1983a). In other animals that swim using pectoral fins, such as the surf perch (Cymatogaster aggregala) and the Humboldt penguin (Spheniscus humboldti), drag during stroking is elevated by an estimated 20–35% (Webb, 1975b; Hui, 1983). Sea lions do not stroke continuously but modulate swimming speed by gliding between flipper beats (Feldkamp, 1985). Because of these factors, it is difficult to estimate the contributions of propulsor drag and values reported here should be viewed as a lower boundary to the drag forces encountered.

Like other swimmers, including fish (Brett, 1965; Webb, 1971, 1975a,b), turtles (Prange, 1976; Butler, Milsom & Woakes, 1984), ducks (Prange & Schmidt-Nielsen, 1970), mink (Williams, 1983), seals (Davis et al. 1985) and humans (Holmer, 1972; Nadel et al. 1974), sea lions exhibit a curvilinear rise in metabolic rate with increasing swimming speed. This is primarily due to the hydrodynamic constraints of the medium. As speed and drag increase, power required for propulsion rises at a rate proportional to U³. Unlike terrestrial animals, which encounter little aerodynamic resistance during running, swimmers must meet these power requirements through an exponential rise in energy consumption.

Despite this, sea lions use less energy for locomotion than terrestrial mammals of similar size. Taylor, Heglund & Maloiy (1982) developed an empirical formula describing the relationship between metabolism and speed for a variety of running animals. When compared to this, SL2 and SL3 expended less energy than a running mammal at all speeds examined. For SL4, swimming was less costly above 1ms⁻¹ (Fig. 6).

These reduced costs can be attributed to several factors. Sea lions are neutrally buoyant in sea water (Table 1). Thus, energy is not expended to support the body during swimming. The importance of this is demonstrated by extrapolating the curve of vs swimming speed (equation 6) to zero velocity. The intercept of this curve accurately predicts the sea lion’s resting metabolic rate. In terrestrial runners the y-intercept consistently overestimates the resting rate, a fact attributed to the cost of maintaining posture. This cost is independent of speed and is an important energetic component at all running velocities (Taylor, 1977). Sea lions, in contrast, rely on the water for support, thereby eliminating this postural component.

Terrestrial runners must also raise and lower their centre of mass during each stride (Heglund, Cavagna & Taylor, 1982). As the centre of mass falls, muscles already under tension are used to slow its progress. This stretches active muscles that, as a result, consume energy but do no positive work (Tucker, 1975; Heglund et al. 1982). In this respect, sea lions again benefit from the support afforded by the medium. It is unlikely that propulsive muscles are stretched while exerting tension during a stroke. Virtually all work performed by the swimming musculature, therefore, contributes directly to the animal’s forward progress and none to raising and lowering the centre of mass.

The minimum cost of transport (CT) is a useful measure for comparing the locomotory system of sea lions with that of other animals, because it depends on the efficiency of the transport process (Schmidt-Nielsen, 1972; Tucker, 1975). For aquatic vertebrates, the minimum CT reflects the importance of swimming and the extent of an animal’s adaptive compromise with other forms of locomotion (Vleck, Gleeson & Bartholomew, 1981). For example, salmonid fish have the lowest CT yet measured (Brett, 1965; Schmidt-Nielsen, 1972). Submerged swimming, poikilo-thermy, neutral buoyancy and an efficient caudal propellor all contribute to this low cost. Swimming reptiles, including turtles (Prange, 1976) and iguanas (Vleck et al. 1981), have costs 1·5–2 times that of similarly sized fish. Surface-swimming ducks (Prange & Schmidt-Nielsen, 1970), mink (Williams, 1983) and muskrats (Fish, 1981, have costs 11–20 times greater. The minimum CT measured for SL2 and SL3 (2·4 J kg⁻¹ m⁻¹) is also 2·5 times that predicted for fish, suggesting that for sea lions swimming is an efficient means of transport. Moreover, SL2 and SL3 have costs equivalent to those reported for phocid seals (Craig & Pasche, 1980; Lavigne et al. 1982; Davis et al. 1985). It seems, therefore, that in pinnipeds foreflipper and rear flipper propulsion are comparable in efficiency.

The aerobic efficiency (N_a) and mechanical efficiency (N_p) of foreflipper locomotion can be calculated using results from the hydrodynamic and metabolic studies. N_a, defined as power output (P_o) divided by the metabolic power input (Webb, 1975a), rose with speed for SL2 and SL3, reaching a maximum of 15 % at 2·7ms⁻¹. The maximum for SL4 was 12% at 2·,6ms⁻¹ (Fig. 7). The efficiency of the foreflippers in converting muscle power to thrust power can be determined in a similar fashion. By subtracting resting metabolic rates from power input values, net power available to the swimming musculature (P_i,_net) ^can be calculated. Power available to the foreflippers will be that supplied by the locomotory muscles.

This approach assumes that energy used by the cardiovascular and respiratory support systems remains constant during exercise. Although a small increase in organ energy consumption undoubtedly occurs at higher work levels, estimates of efficiency based on this assumption will err on the conservative side. It is also assumed that a reduction in efficiency and the effects of a lower propulsor drag are negligible at the higher work rates but lower absolute swimming speeds created by the addition of drag cups. Because foreflippers create thrust independently of the body, the estimate of equivalent speeds with extra drag should be accurate given this assumption.

These efficiencies are relatively high, compared with those of other swimmers. The semi-aquatic mink has an N_a of only 1-8% (Williams, 1983). Similarly, the muskrat has an N_a of 4·6% with an N_p of 33% (Fish, 1984). Prange (1976) has shown that the N_a for green sea turtles is of the order of 0·09, while the surf perch Cymatogaster, which uses its pectoral fin for swimming, has an N_a of 0·12–0·13 with an N_p of 0·60–0·65 (Webb, 1975b). Fish that swim in the carangiform mode have the highest efficiencies reported. The mechanical efficiencies of rainbow trout and sockeye salmon are 0-7 and 0-9 with N_a values of 0·15 and 0·22, respectively (Webb, 1975b)

The shape of the propulsive appendages and the kinematics of thrust generation determine, in large part, these efficiencies (Webb, 1975a,b;Blake, 1983a; Fish, 1984). Mink and muskrats swim by paddling. Thrust is generated during the power phase of the stroke, but recovery introduces substantial drag as the appendages are brought forward to their initial position (Williams, 1983; Fish, 1984). Like sea turtles (Walker, 1971) and the surf perch (Webb, 1975b), sea lions use a lift-based method of propulsion. Thrust is generated in both the power and recovery phases of the stroke cycle (English, 1976; Feldkamp, 1985; Godfrey, 1985). The sea lion’s mechanical efficiency is undoubtedly improved by this style of swimming. It is further improved by the design of the foreflippers. They are dorsoventrally compressed and hydrofoil-shaped, with an FR that ranges from 5·4 at the tip to 3·2 at the base (Feldkamp, 1985). This shape reduces pressure drag and improves lift. In addition, they have a comparatively high aspect ratio of 7·9 (Table 1). A long, narrow hydrofoil will diminish induced drag as lift is created during a stroke (Vogel, 1981 ; Alexander, 1983). It can, therefore, produce forwardly directed lift more efficiently because it imparts a small degree of downward momentum to a relatively large mass of water (Vogel, 1981).

In conclusion, swimming is a comparatively inexpensive means of transport in the California sea lion. Its overall shape approximates to an ideally streamlined object with a noticeable absence of any surface projections that would tend to increase drag and elevate the power requirements for movement. During swimming, the body is supported by the water, negating the postural costs associated with terrestrial locomotion and reducing the amount of negative work performed by the swimming musculature. The foreflippers hydrodynamic shape, high aspect ratio and continuous generation of propulsive forces all contribute to the high mechanical efficiency of these appendages. As demonstrated by the low transport costs, adaptations for swimming have not been seriously compromised by the need for mobility on land, and enable sea lions efficiently to exploit the marine environment.

This work was conducted as part of a doctoral dissertation project, supported by NIH grant USPHS HL17731 to Dr G. L. Kooyman. I would like to extend my thanks to Dr G. L. Kooyman for his guidance and support. The assistance of Drs T. M. Williams and R. W. Davis and Mr P. Thorson in all phases of this project is gratefully acknowledged. I also thank Drs F. White, J. Graham, P. Webb and my colleagues at Long Marine Laboratory for useful discussion and valuable comments. Special thanks to J. Feldkamp for help with the figures.