Aspects of Shark Swimming Performance Determined Using a Large Water Tunnel

Because of the size of most water-tunnel respirometers, studies of fish performance and physiology at controlled swimming speeds have usually been limited to specimens smaller than about 40 cm (Yates, 1983). For this same reason the relatively large size of most sharks has precluded investigations of their swimming performance or measurements of their activity metabolism at controlled velocity (Bone, 1988). The few studies of the cardiovascular physiology and metabolism of sharks during exercise were made without benefit of control of either swimming duration or speed (Piiper et al. 1977; Brett and Blackburn, 1978; Bushnell et al. 1982, 1989; Gruber, 1986; Scharold and Gruber, 1990).

Recent development of a large seagoing water-tunnel respirometer at Scripps Institution of Oceanography (SIO) has enabled examination of various aspects of shark swimming performance including O₂ consumption rate and critical swimming velocity (U_crit). The objectives of this paper are to describe the SIO water tunnel, to present data comparing aspects of the swimming performance of three shark species, and to compare the swimming capacity of sharks with certain continuous-swimming teleosts. Although limited in scope, investigations of shark swimming performance conducted to date (Piiper et al. 1977; Brett and Blackbum, 1978; Bushnell et al. 1982) suggest that these animals have low metabolic rates compared with bony fishes and that they have little or no capacity for sustained aerobic swimming. To examine this and to quantify some aspects of the relationship between shark body shape and swimming proficiency proposed by Thomson and Simanek (1977), we carried out detailed studies with the California leopard shark Triakis semifasciata, as well as selected observations on a single lemon shark Negaprion brevirostris that was shipped to La Jolla from Florida, and specimens of the mako shark Isurus oxyrinchus captured and experimented with at sea. By obtaining metabolic data for the lemon shark, a species adapted for continuous swimming and the subject of recent study (Gruber, 1986; Bushnell et al. 1982, 1989; Scharold and Gruber, 1990), we have verified that the SIO water tunnel can be used to make measurements in agreement with other published data, although at faster swimming speeds. In addition, our paper reports the first data on the U_crit and ram gill ventilation of the leopard shark, a coastal, bottomfeeding species, and the first measurements of swimming and tail-beat frequency of the mako shark, a pelagic species and member of the family Lamnidae which is specialized for continuous, thunniform swimming.

The SIO water tunnel (Fig. 1), is a large (24001) Brett-type recirculating treadmill mounted on a 3.3 m ×5.3 m pallet. The unit can be used in the laboratory or taken to sea. Water flow through the instrument is driven by two 12-inch (30.5 cm) propellers mounted on a shaft connected by a flexible coupling to a 30 kW variable-speed motor operated at 440 V. Circulating water first passes into the diffuser-contraction section which reduces large-scale turbulence, then through a 7 cm long honeycomb collimator (0.32 cm cell diameter) to provide microturbulent flow through the working section. The working section is 200 cm × 51 cm × 42 cm (length x width x height) and both its outer-side wall and removable lid are constructed of clear Lucite to facilitate observation of the swimming fish (Fig. 2). The lid also contains a removable standpipe for passing monitoring wires and cannulae from the working section to the outside. Two small ports on the side wall permit water shunting for analysis of chemical properties and respiratory gases. A metal grating at the downstream end of the working section prevents the fish from being swept back into the 30.5 cm polyvinylchloride pipe leading to the propellers. A 0.75 kW pump draws some of the system water through a six-element pool filter and a 2.9kWh chiller (Fig. 1).

Working-section velocity calibrations were made with a flow meter (General Oceanics Inc. model 2035-mk III) mounted mid-channel and mid-depth and halfway back in the section. In the present configuration, the maximum working section velocity is 2.0 ms⁻¹, but most experimental data have been obtained at speeds up to 1.3 ms⁻¹. The motions of attached threads and injected dye streams indicate a uniform field of laminar flow through the first half of the working section. Observation of particle motion shows flow along the edges to be smooth but slower than in mid-channel.

Leopard sharks were captured in Elkhorn Slough, Moss Landing, California, USA, and transported by truck to holding facilities at SIO. Sharks were held in four 700-1 circular fibreglass tanks continually supplied with fresh, aerated sea water and maintained at ambient temperature (14–24°C) and photoperiod. All sharks were regularly fed either squid or mackerel but fasted for 1 week prior to experiments.

A 70cm long lemon shark was captured by dip net in the Marquesas Keys, Florida, USA, and transported to SIO by air. This fish was held in a heated tank (21–26°C) which was aerated and supplied with a slow but continuous flow of fresh sea water, and was fed mackerel or squid several times weekly.

Mako sharks were captured by surface long line off the coast of southern California during seagoing operations aboard the SIO R/V R. G. Sproul in July, 1987. Captured sharks were placed in a holding tank or directly into the working section and held there, without feeding, for the duration of the study.

Work with makos was done opportunistically as they were collected at sea and placed in the water tunnel. Laboratory tests involving leopard and lemon sharks permitted repeated use of the same specimens in the water tunnel; this allowed them to become accustomed to the confines of the working section and to learn to swim within its boundaries. The mass (kg) and length (L in cm) of all fish studied were determined upon completion of water-tunnel tests. Unless indicated otherwise, all tests and observations were performed at specified ambient temperatures and at or near O₂ saturation levels.

Estimates of the for each of the three shark species during swimming were obtained by measuring the rate of decline in the O₂ content of water in the respirometer during periods when fish were swimming steadily and factors such as temperature did not fluctuate. Detailed procedures for this are contained in Graham et al. (1989) and Scharold et al. (1989). Measurements with Triakis and Negaprion were made on individuals that were post-absorptive and habituated to the water tunnel by replicate testing and were carried out between 3 and 7 h after handling. Isurus respirometry (see Results) was performed on a freshly collected fish over a period extending from several to nearly 48h after capture. Water O₂ tension was monitored with a temperature-compensated probe (YSI5450/5758) and meter (YSI 54A) interfaced to a computerized data-logging system (Keithley DAS series 500). The probe was calibrated in air and water before each run and was mounted in a flow-through cuvette connected to the working section by a peristaltic pump. All data were corrected for background respiration determined by blank runs soon after removal of the test fish.

Tail-beat and gill-ventilation frequencies of swimming fish were measured in a few fish using impedance monitors and whenever possible in non-instrumented fish by making replicated counts over 1 min.

Table 1 shows U_crit values, expressed in terms of both absolute (cms⁻¹) and length-specific (Ls⁻¹) speed, for 18 leopard sharks and one lemon shark. The ranges of water temperature and O₂ level determined during each test are also indicated. Replicate runs on two leopard sharks and the lemon shark did not demonstrate any temporal effects and the cooler water temperature used in some of the lemon shark tests did not affect its U_crit. Fig. 3 shows the significant (P<0.05) inverse relationship between U_crit (Ls⁻¹) and body length for leopard sharks and data for the lemon shark. For comparative purposes, U_crit data for the sockeye salmon (Oncorhynchus nerka, Brett and Glass, 1973), ranging in length from 5.5 to 61.4 cm, are also shown in Fig. 3. It is apparent that the U_crit of leopard sharks between 50 and 60cm in length is about 0.7Ls⁻¹ less than that of the sockeye.

Fig. 4 shows tail-beat frequency (TBF) in relation to absolute velocity for 15 leopard sharks separated on the basis of body length into three groups. This figure shows that at comparable speeds smaller (30–60 cm and 61–90 cm) leopard sharks have a higher TBF than do larger sharks and that TBF tends to reach a maximum at higher velocity, confirming earlier observations of Scharold et al. (1989). Fig. 5 permits comparison of the 61–90 cm leopard shark TBF data set (expressed as a regression of TBF on relative speed) with data and the regression line for the lemon shark. In both sharks, the increase in TBF with speed, while linear over most of the velocity range, tended to become asymptotic at maximum relative speed.

TBF data for three specimens of Isurus (82–125 cm, 4–16 kg) are shown in Fig. 6 along with the TBF-velocity relationship for the smallest (82cm) of these. For comparison, the lines for the 70cm Negaprion and the 61–90cm group of Triakis (both from Fig. 5) have been added, and a body shape profile of each species is shown next to its regression line (see Discussion). The Isurus fine has a lower slope than those of Triakis and Negaprion, but the observed difference is significant only at the 10% level.

Gill ventilation frequency was quantified for 10 leopard sharks (52–121 cm). With increased swimming velocity, sharks 71 cm and longer (N=6) were observed to switch from continuous, active gill ventilation to a non-cyclic and passive ram ventilation. This transition was most pronounced in the three largest sharks, which reduced their ventilations to between 0 and 0.02 Hz at swimming speeds inversely related to body length (Fig. 7). In three smaller sharks (71–77 cm), ventilation also became reduced at speeds between 35 and 60cms⁻¹ (mean=43cms⁻¹), but frequencies of less than 0.2 Hz were not observed. The four smallest sharks were not observed to ram ventilate.

Fig. 8 shows the swimming velocity vs data obtained for Triakis (first published in Scharold et al. 1989), Negaprion and Isurus in the SIO water tunnel in relation to two vs swimming velocity regressions recently determined for Negaprion (Bushnell et al. 1989; Scharold and Gruber, 1990). The lowest was measured for T. semifasciata [mean resting mean during swimming at an average maximum velocity of 0.93 Ls⁻¹ =167 mg kg⁻¹h⁻¹ (Scharold et al. 1989)]. Combining all metabolic data determined for the 70 cm lemon shark (Negaprion) in this study yields a of 318±96mgkg⁻¹h⁻¹ (X̄±2S.E., range 139–627, N=9) at 1–1.3 Ls⁻¹. Fig. 8 indicates how our study expands the velocity range over which metabolic data for this species are now available. Also, our estimates for this fish at different temperature combinations (22°C and 25 °C) and swimming speed lie reasonably in the extrapolated path of regression lines determined by other investigators.

At sea an 82 cm, 3.9 kg mako shark swam in the water tunnel for 96h (3–7 July, 1987). During a 41-h period that began 4h after capture and placement in the water tunnel, 36 h of metabolic data were obtained (Table 2). During these measurements, the shark’s speed ranged between 0.2 and 0.5 Ls⁻¹ and temperature from 16 to 20°C. Table 2 shows little effect of ‘time since capture’ on metabolism and combining all estimates yields a mean swimming ⁠, of 369±llmgkg⁻¹h⁻¹. The of this mako varied steeply over the range of velocities tested (Fig. 8). Because numerous measurements were made at speeds of 0.26 and 0.30Ls⁻¹ (Table 2), mean (±S.E.) ⁠, values are shown for these speeds in Fig. 8.

As reviewed by Yates (1983), nearly all studies of fish swimming performance (excluding tunas, Graham et al. 1989) have been made with specimens smaller than 40 cm in length. The present study shows that the SIO tunnel will allow work with somewhat larger fishes. For example, the cross-sectional area of the 82cm mako shark (95 cm²) amounted to only 4.5% that of the working section (2100 cm²) and thus was too small (less than 10%) to require a solid blocking correction factor. The height (42 cm) and width (51 cm) of the working section also exceed the dimensions of the large fish. The 82cm mako, for example, had a pectoral span of 31 cm and a maximum vertical expanse (at the dorsal fin) of 19 cm. Thus, and provided that swimming occurs in the centre of the section (which was not always the case), wall effects are also not a continuous problem for larger fishes in the SIO tunnel.

Studies with larger animals also reduce some surgical and instrumentation problems and permit investigations of aspects of fish metabolic performance and physiology (for example, close-order regional electromyography) previously not feasible. Working-section features, such as the standpipe and transparent lid and side wall, permit physiological and kinematic studies, and data presented here show that the increase in water-tunnel volume caused by inclusion of the specially designed diffuser-contraction section has not hampered its utility as a respirometer. Finally, because it can be taken to sea, the SIO water tunnel enables firsttime research with pelagic fishes, such as the mako shark, that could not otherwise be studied. Pelagic fishes are generally too large for other swimming tunnel systems and do not survive well enough in captivity to permit study in a land-based laboratory.

Interspecific U_crit comparisons must be made cautiously because values are affected by fish body size, by ambient conditions, and by the combination of duration and velocity increment used in their determination (Beamish, 1978). The U_crit of the sockeye was chosen for comparison with Triakis because fishes from both studies overlap in body size. The inverse relationship between U_crit and body size (Fig. 3) seen for both species probably reflects the disproportionate scaling of factors affecting performance (e.g. cardiac output, red muscle amounts) and factors influencing surface drag (e.g. body size and shape, surface area and fin placement) in larger fishes (Brett and Glass, 1973; Beamish, 1978; Yates, 1983; Bone, 1988).

Interpretation of the actual differences between Triakis and Oncorhynchus is complicated by differences in experimental time increment used in the two studies (30min tests for Triakis vs 60min for sockeye). Two facts, however, persuade us that the time effect is sufficiently small to warrant this comparison. First, Beamish (1978) found the U_crit of another salmonid (the rainbow trout, Salmo gairdneri) was not affected by test durations ranging between 20 and 60 min. Also, 30-min U_crit determinations for two marine teleosts by Freadman (1979) (3.3 Ls⁻¹ for 24–29 cm striped bass, Morone saxatilus’, 4.6 Ls⁻¹ for 18–25 cm bluefish, Pomatomus saltatrix), while higher than that of Triakis, are very close to the U_crit expected for a comparably sized Oncorhynchus on the basis of Fig. 3.

Recognizing that U_crit is an index of aerobic swimming capacity, the lower U_crit of Triakis relative to that of Oncorhynchus may be partly related to anatomical and physiological differences in red muscle function in sharks and teleosts (Bone, 1988). Both sharks and teleosts have similar amounts of red muscle (Bone, 1978) and the aerobic capacity of this tissue seems comparable in both groups (Bone, 1988; Dickson et al. 1988). An important difference affecting sustained performance by shark red muscle may be the strong division of labour that exists between the red and white muscle in this group. Electromyography has demonstrated that the focally innervated white fibres found in sharks are active only during burst swimming (Bone, 1966). This differs from many higher teleosts in which the white fibres are multiply innervated and can work simultaneously with red fibres in powering sustained swimming (Bone, 1966, 1978, 1988). Thus, while both sharks and teleosts have similar amounts of red muscle, the ability of teleosts to recruit different groups of white muscle fibres during sustained swimming may afford them a higher U_crit.

Differences in the swimming capacity of teleosts and sharks have doubtlessly influenced, and been influenced by, their life histories. Small teleosts, for example, can attain high relative speeds (Fig. 3 shows that a 10-cm long sockeye salmon has a U_crit between 6 and 8Ls⁻¹), which is thought to be important in escaping predators (Brett, 1964; Beamish, 1978). The shape of the U_crit function of Triakis also extrapolates to a higher value with smaller size, but the increase is much less than for the sockeye. Because sharks are often 20 cm or longer at the time of birth or hatching, small sharks would require less of the predator-avoiding capability, afforded by a high U_crit, than do small teleosts.

Examination of TBF in relation to velocity and body size (Fig. 4) shows that smaller (30–60 cm) Triakis had a higher TBF and attained a greater relative speed than did either group of larger fish. The finding that maximal TBF is sizedependent, although not previously documented for sharks, is similar to the situation in teleosts (Bainbridge, 1958). Previous studies with both teleosts and elasmobranchs reveal a generally linear relationship between TBF and velocity, although the shark data are limited (Hunter and Zweifel, 1971; Webb and Keyes, 1982). Our data show that the TBF of Triakis and Negaprion tended to plateau near U_crit (Fig-4)- One interpretation of this could be that exhaustion curtailed the sharks’ ability to sustain a high TBF. However, we and Scharold et al. (1989) found similar relationships between TBF and velocity for Triakis swum for shorter periods. This again suggests that the slow contraction velocity of red muscle and the time required for red fibres to recover their resting length before contracting may limit maximum sustainable TBF.

The maximum TBF recorded for sharks in this study is lower than that of (generally smaller) teleosts at similar velocities. This is consistent with differences in myotomal organization described above and supports the idea that, in the absence of white fibre augmentation, the maximum contraction-relaxation cycle of red muscle may limit TBF and therefore velocity. In addition, the tendency for TBF to reach a maximum suggests that other factors affecting swimming thrust, such as propulsive wavelength and caudal amplitude, may be important at near maximal sustained velocities and as the faster-contracting white fibres are recruited. Kinematic and electromyographic studies are needed to determine this. Webb and Keyes (1982) observed a specimen of Carcharhinus melanopterus to change both wavelength and amplitude when (burst) swimming at 3.9 Ls⁻¹.

Interspecific comparisons by Webb and Keyes (1982) indicate that sharks adapted for sustained swimming have a lower relative TBF and greater stride length (distance travelled per tail beat). This is consistent with our findings for Triakis, Negaprion and Isurus, members respectively of groups 3, 2 and 1 of the Thomson and Simanek (1977) classification of body form in relation to swimming proficiency. Although the slope of the TBF-velocity function presented for Isurus in Fig. 6 differs from that of the other two genera only at the 10% level of significance, this figure does suggest that there might be functional differences among these genera (and groups 1, 2 and 3) in factors affecting propulsion and drag, such as body profile and caudal fin shape and surface areas. From its body shape and caudal morphology, Triakis (Fig. 6) appears better adapted for unsteady swimming and periodic resting on the bottom, which is consistent with its normal behaviour in shallow coastal waters. Also, the implication of Fig. 6 that Triakis has a higher TBF than either Negaprion or Isurus at the same relative speed suggests either a greater thrust requirement or less thrust generation per tail beat. By contrast, the morphologies of Negaprion and especially Isurus are more suited to cruising. Additional data for both Negaprion and Isurus are needed to quantify factors affecting drag and thrust generation (e.g. body shape and streamlining, surface area, and caudal fin shape and area) and to determine whether functional morphological features such as stride length and caudal amplitude also affect aerobic swimming efficiency.

Our findings for Triakis agree with observations on ram ventilation in a similarly sized group of T. scyllia by Kabasawa and Clark (1974–1975). Ram ventilation, a respiratory mode in which the forward movement of the fish is sufficient to develop the pressure head necessary to force water through the gills, is commonly seen among teleosts adapted for cruise swimming and is thought to reduce metabolic costs by as much as 15% (Roberts, 1975, 1978; Freadman, 1979). The range of speeds (27–60 cm s⁻¹) over which the larger Triakis adopted ram ventilation agrees with the range seen for two teleosts (Pomatomus and Morone, 33–65 cms⁻¹; Roberts, 1975; Freadman, 1979). However, whereas Freadman (1979) found only a slight effect of body size on the ram-transition velocity of Pomatomus and Morone, our observations with Triakis ranging from 50 to 120 cm in length revealed that ram ventilation did not occur in smaller (less than 70 cm) sharks and that the speed at which it began in larger sharks was inversely related to body length (Fig. 7). Thus, ram ventilation in Triakis is restricted to larger specimens and is absent in fish comparable in size to most of the teleosts (15–35 cm) for which this phenomenon has been documented. Additional studies are needed to determine factors affecting the ram-transition velocity in differently sized Triakis and to determine why this activity occurs in small teleosts but not in small Triakis.

Including data reported here, swimming velocity vs metabolism estimates are now available for five shark species. Prior to our work, only the of Squalus, Scyliorhinus and Negaprion had been measured, but in no cases had swimming speed been controlled. Brett and Blackburn (1978) could not induce Squalus acanthias to swim in a respirometer but measured at different activity states, reporting the following mean rates (2 kg fish at 10°C): resting 32, routine 49 and active 88 mg O₂ kg⁻¹h⁻¹. Piiper et al. (1977) found that the resting of Scyliorhinus stellaris (2.5 kg at 19°C) was 92 mg O₂kg⁻¹h⁻¹ and increased to 162 mg kg⁻¹ h⁻¹ during spontaneous swimming at 0.27 Ls⁻¹ for 1-22 min. They also reported that fish swimming for 10 min or longer acquired an O₂ debt.

In a review based primarily on data for docile species, Brett and Blackbum (1978) concluded that the metabolic rates of elasmobranchs are ‘among the lowest of all fishes’, but allowed that active sharks probably had higher metabolic rates. This is now verified in work with Negaprion brevirostris (Gruber, 1986; Scharold and Gruber, 1990; Bushnell et al. 1989). Our study with the 70-cm Negaprion additionally demonstrates that this species can sustain swimming speeds above ILs⁻¹ and metabolic estimates at this speed (mean ⁠) lie in the extrapolated path of regression values obtained at slower speeds (Fig. 8). The high metabolic rate of swimming elasmobranchs is further documented by our data for I. oxyrinchus. Although our data are for only one specimen, they span nearly 48 h and are the first data reported for a lamnid shark, as well as the highest. in relation to velocity yet measured for an elasmobranch (mean at 0.2–0.5 As⁻¹).

Fig. 9 shows the velocity data presently available for five different shark species in relation to the resting vs active range (40–106mgkg⁻¹h⁻¹) set by Brett and Blackbum (1978) and in relation to data for four cruise-adapted teleosts. While comparisons among these data are doubtlessly complicated by numerous variables such as differences in water temperature and in thermal adaptation and acclimation states and differences in body mass and in relative activity level, it is apparent that most of these metabolic values are either at the high end of, or above, the Brett and Blackburn (1978) range. It can also be seen that the values of the mako, the lemon and the leopard sharks line up in the same order as ascribed to the form of their bodies for proficient swimming (Thomson and Simanek, 1977). Also, the of the mako approaches that of tunas (skipjack and albacore), which is not surprising in view of the evolutionary convergence in body form and swimming specializations found in the families Scombridae and Lamnidae, including the presence of endothermy in both the mako and tunas (Graham, 1983). Fig. 9 also suggest that sharks have a smaller cruising velocity range than teleosts. This may be an artefact of the relatively few shark studies to date, but there are few accounts of sharks cruising much faster than 1L s⁻¹, which may reflect both their optimal velocity and the limitations on cruising imposed by the division of labour between red and white myotomal muscle.

This work was supported by NSF DCB84-16852, DCB87-18449 and DMB85-00261. Ship time was provided by the University of California, UC-10, 1987. We thank the captain and crew of R/V Robert Gordon Sproul for assistance and support. Appreciation is expressed to R. Dotson, J. Scharold, T. Okey, D. Schonick, S. Sutton, D. Ticaett, A. Escandon, E. Bliase and D. Ward for their technical assistance at sea and in the laboratory. We thank Drs John Hunter, Sergey Kashin and Paul Webb for their comments and suggestions on early manuscript drafts.