Maximum Sustainable Speeds and Cost of Swimming in Juvenile Kawakawa Tuna (Euthynnus Affinis) and Chub Mackerel (Scomber Japonicus)

Tunas and mackerels, members of the teleost fish family Scombridae, are epipelagic predators capable of attaining high swimming speeds as well as undertaking long-distance migrations (for reviews, see Beamish, 1978; Magnuson, 1978; Joseph et al., 1988). All members of this family possess anatomical characteristics, such as a streamlined body shape and a high-aspect-ratio caudal propeller, that reduce drag or increase thrust production (Webb, 1975; Collette, 1978; Magnuson, 1978; Dewar and Graham, 1994). Scombrids also have caudal keels and finlets along their narrow caudal peduncles that are thought to aid in minimizing the turbulence encountered by the caudal fin as it moves laterally during swimming (Aleev, 1963; Collette, 1978; Magnuson, 1978). Although the tunas share these morphological traits with other scombrids, they alone can conserve metabolically derived heat to maintain the temperature of the slow-twitch, oxidative locomotor muscle significantly above that of the water and, thus, gain some independence from environmental temperature (Carey et al., 1971; Graham, 1973). Thus, tunas are endothermic and are considered to be regional heterotherms. The tunas constitute a monophyletic clade within the family Scombridae, and all taxa of the sister groups, the bonitos, mackerels and Spanish mackerels, are ectothermic (Carey et al., 1971; Collette, 1978; Block et al., 1993; Graham and Dickson, 2000). Thus, the Scombridae provides an ideal system with which the selective advantages and the evolution of endothermy can be investigated.

Several hypotheses have been proposed to explain why tunas are endothermic. One hypothesis suggests that tunas maintain elevated temperatures to slow the rates of temperature change in their tissues when they swim into or spend brief times in cold water, thereby sustaining high activity levels when subjected to reduced water temperatures (Graham, 1975; Neill et al., 1976; Stevens and Neill, 1978; Brill et al., 1999). It has also been hypothesized that tunas evolved their endothermic capabilities to expand their ranges into cooler waters (Block et al., 1993; Graham and Dickson, 2000). Alternatively, endothermy may have evolved to allow tunas to increase their rates of certain metabolic processes, including growth, digestion and processing of waste products such as lactic acid (Carey and Teal, 1966; Stevens and Carey, 1981; Stevens and McLeese, 1984; Brill, 1996). A final hypothesis is that warming the aerobic myotomal muscle tissue increases swimming performance by increasing muscle contraction velocity, power output and swimming efficiency (Carey et al., 1971; Graham, 1975; Brill and Dizon, 1979; Stevens and Carey, 1981; Johnston and Brill, 1984; Dewar and Graham, 1994; Altringham and Block, 1997). This could increase the maximum sustainable swimming speeds of tunas and enable them to be more efficient long-distance swimmers.

Although the swimming energetics of tunas larger than 320 mm in fork length (FL) have been described (Gooding et al., 1981; Graham et al., 1989; Dewar and Graham, 1994), the last hypothesis remains untested because there have been no comparisons between tunas and closely related ectothermic species that clearly identify the locomotor advantages associated with endothermy. Previous studies that have examined the energetics of tunas and then compared the results with those from an ectotherm of similar size have had to use data from salmonids for this comparison. Although a salmonid may offer a sized-matched comparison, the conclusions that may be drawn are limited because the two groups of fishes are so distantly related and because the studies were conducted at significantly different temperatures. The present study used juveniles (110–250 mm FL) of the kawakawa tuna Euthynnus affinis and the chub mackerel Scomber japonicus to test the hypothesis that aerobic swimming performance is greater in the tunas than in their ectothermic relatives. It is also the first study successfully to measure maximum sustainable swimming speeds of tunas under controlled conditions. Tunas of this size range represent those above and below the hypothesized minimum size for endothermy (207 mm FL) based on studies of the black skipjack tuna Euthynnus lineatus (Dickson, 1994). Thus, the present study also provides an intraspecific comparison of swimming performance in tunas that can and cannot elevate their muscle temperature.

Juvenile kawakawa Euthynnus affinis (Cantor, 1849), ranging in size from 110 to 250 mm fork length (FL), and one yellowfin tuna Thunnus albacares (Bonnaterre, 1788) were captured by traditional hook- and-line methods in July and August 1997 around fish-aggregating devices off the leeward side of the island of Oahu, Hawaii. The fish were immediately placed in a circular 110 l bait tank mounted to the transom of a sportfishing vessel. The water in the tank was maintained at the same ambient sea surface temperature at which the fish were caught (24±2 °C) and was supersaturated with oxygen. The fish were then transported to shore, transferred to a 380 l circular tank containing oxygenated sea water mounted in the bed of a pickup truck, and transported to the National Marine Fisheries Service (NMFS) Kewalo Basin Research Facility, Honolulu, Hawaii. At the NMFS laboratory, the fish were maintained for 2–8 weeks in an 8 m diameter above-ground plastic-lined outdoor tank supplied with continuously flowing sea water at an ambient temperature of 24±2 °C and salinity of 33 ‰, and were subjected to a natural photoperiod. Initially, the tunas were fed live red-tailed guppies (Poeciliidae) until they began to eat chopped fish. They were then fed three times a day on an assortment of chopped fish [jack (Decapterus macarellus and Selar crumenophthalmus) and skipjack tuna (Katsuwonus pelamis)]. The amount of food consumed at each feeding was estimated to be approximately 10 % of body mass. Each tuna was feeding regularly in the tank for a minimum of 10 days before being chosen for an experimental trial.

Chub mackerel Scomber japonicus (Houttuyn, 1782) ranging in size from 133 to 250 mm FL, were collected in the coastal waters of southern California, USA, near Dana Point Harbor, between May and August 1998. The fish were transported from Dana Point Harbor in an 80 l round bait tank and were maintained in a 1.6 m diameter, 1500 l indoor fiberglass tank at California State University Fullerton, in water with a salinity of 34–35 ‰ and at a temperature of 24±2 °C. The mackerel were fed chopped fish [northern anchovy (Engraulis mordax) and Pacific sardine (Sardinops sagax)] and squid once a day, in an amount approximating 10 % of their body mass. The mackerel were subjected to a 12 h:12 h light:dark cycle and were feeding regularly in the tank prior to selection for any experimental trial. For all species, no feeding was performed for at least 12 h prior to selection for the respirometer.

A Brett-type swimming tunnel respirometer. was used to measure the rate of oxygen consumption rate of the fish as a function of swimming speed (Fig. 1). The respirometer, similar to but smaller than that used by Dewar and Graham (1994), is a swimming tunnel that constantly recirculates 35 l of water. A variable-speed motor propels the water, so that a fish maintaining its position in the center of the chamber is swimming at the same speed as the flowing water. A General Oceanics mechanical flow meter (model 2030 R) mounted in the center of the chamber was used to calibrate respirometer speeds prior to any respirometry trials and each time the chamber was moved or disassembled. Flow characteristics were also assessed visually by observing dye flow patterns at various speeds without fish swimming in the respirometer. Corrections for solid blocking effects were not necessary because the experimental subjects had cross-sectional areas that were less than 10 % of the cross-sectional area of the chamber (Brett, 1964; Webb, 1971).

The respirometer system can be used in both an open and a closed mode. When open, the water in the respirometer is circulated to and from a reservoir, where. it can be oxygenated to the desired level. When closed, the of the fish can be measured. Water oxygen concentration (±0.3 mg O₂ l⁻¹) was recorded continuously by a Yellow Springs Instruments (YSI) less than 120 mm FL and 10 cm s⁻¹ for larger fish. was polarographic oxygen electrode (YSI model 5750) connected to an oxygen meter (YSI model.58), which then relayed the signal to a chart recorder. The of the fish at each speed was calculated from the rate of change in water oxygen concentration. Prior to each experimental run, the YSI electrode membrane was replaced, and the probe was calibrated in air.

For all species, each experimental run began with the introduction of a fish into the respirometer. This was accomplished by hooking the fish with a baited barbless hook and transferring it to the chamber in a plastic pail liner filled with oxygenated sea water. The respirometer was then covered with a black cloth to minimize outside disturbance to the fish. Each fish underwent an initial 4 h recovery period at a speed of approximately 20 cm s⁻¹ during which the oxygen level was maintained at approximately 200 % saturation for the first 3 h and was then allowed to decline to 100 % saturation over a further 1 h. The initial high O₂ concentration was used to ensure that the fish would recover from capture and transfer to the respirometer while at the initially low water speeds. Dewar and Graham (1994) no.ted a marked change in tuna behavior and a decrease in after 2 h inside a respirometer; furthermore, it has been found that skipjack tuna can clear lactate produced during exhaustive exercise within 2 h (Perry et al., 1985; Arthur et al., 1992).

To ensure adequate time for these scombrids to adjust to the chamber, to process. any lactate produced during the transfer procedure and for to drop and stabilize, a 4 h recovery period was chosen. Each experimental run consisted of a 30 min period at each of the designated speeds up to the maximum sustainable speed. The rate at which the speed was increased depended on the size of the fish: 7.5 cm s⁻¹ for fish measured during the second 10 min of each 30 min period at each speed. During the final 10 min period, the respirometer water oxygen content was brought back to the original 100 % saturation level by allowing water to flow to and from the reservoir. The level of oxygen saturation in the respirometer never dropped below 70 % for any of the trials. All experiments were conducted using filtered sea water maintained at an ambient temperature of 24±1 °C with an in-line heater and chiller. After each experimental run, the fish was removed, and the respirometer was re-sealed for the determination of t.he background respiration rate, which. was subtracted from all measurements to determine fish at each speed.

The maximum swimming speed that the fish was able to maintain for a complete 30 min period while swimming using a gait characterized by steady, continuous tail beats (U_max), assumed to be powered by the slow-twitch, oxidative myotomal muscle (Videler, 1993; Rome, 1995; Webb, 1998), was determined by direct observation. The criterion for defining U_max was when the fish shifted from a steady tail beat to a ‘burst- and-glide’ swimming mode three times within 30 s. When the fish reached this point, it changed positions in the chamber, drifting back towards the back grating, brushing its tail against the grating and bursting forwards or swimming from side to side near the back of the chamber. Behavioral observations from the posterior of the fish, viewed through a hole cut into the black cloth covering the respirometer, were recorded throughout the experiment. Subsequent analysis of the versus speed relationship for each individual was used to confirm a continuous increase in with velocity up to Umax.

For some individuals, the decreased slightly or stayed the same at the highest speeds, indicating that the fish was not actually swimming at the recorded water speed. This was confirmed by checking the observation records, which indicated that, at the highest motor speeds, these individuals swam from one side of the chamber to the other or that they avoided the center of the chamber and positioned themselves close to one side, but did not yet exhibit burst- and-glide behavior. When the fish is positioned against the sides of the chamber, interactions between the fish and the walls of the chamber produce frictional forces that can reduce the speed of the water flowing near the fish (Webb, 1993). For this reason, the fish was able to complete a period at a higher motor speed without consuming any more oxygen than it had consumed at the previous speed, because it was not actually swimming at the higher speed. In such cases, we did not attribute the leveling off of to the fish having reached its anaerobic threshold because these active fishes are unlikely to be able to power continuous swimming for as long as 30 min with anaerobic metabolism. Moreover, we would have expected the fish to switch to a burst- and-glide gait if they recruited the fast, glycolytic muscle fibers (Rome, 1995; Webb, 1998), but the behavioral observations indicated that all individuals swam by beating their tails continuously at those speeds. For these individuals, U_max was determ. ined to be the highest speed at which the fish swam before leveled off or declined. For allother individuals, the highest speed at which the fish swam for a full 30 min period was considered to be the U_max.

The net cost of transport (COT_net, in J km⁻¹ kg⁻¹), or incremental cost of swimming, was used to quantify swimming efficiency; the lower the COT_net value, the more efficient the fish. COT_net was calculated by .first subtracting the b.ackground respiration rate from each ⁠, then plotting versus swimming speed for each fish; the slope of the best-fitting linear regression of the resulting graph is proportional to the mean net cost of transport for the fish. The slope units (mg O₂ min⁻¹ cm⁻¹ s) were converted to J km⁻¹ kg⁻¹ by multiplying by an oxycalorific equivalent (3.24 cal mg⁻¹ O₂; Beamish, 1978) and converting cal to J (4.18 J cal⁻¹).

Muscle temperature measurements were taken for both species with a digital thermocouple thermometer (Omega Engineering, model HH-72 T). After the respirometry trial and subsequent videotaping for a companion kinematic study (Donley and Dickson, 2000), which took approximately 30 min, the motor speed was reduced to approximately 1 body length s⁻¹, the chamber lid was removed, and the fish was quickly grasped around the opercular region. A thermocouple embedded in a 21-gauge hypodermic needle was immediately inserted into the deep red muscle of the kawakawa, penetrating to the vertebral column, or into the superficial red muscle of the chub mackerel, at a series of locations along the length of the body, and temperatures were recorded. The temperature of the water was then taken with the same thermocouple, and the difference between the water and the maximum red muscle temperature recorded was defined as the maximum muscle temperature elevation (T_x).

All procedures were conducted with the approval of the CSUF Institutional Animal Care and Use Committee.

The U_max data for both species were first examined for normality in Minitab (version 10.5), and it was determined that the data did not need to be transformed prior to analysis. The U_max data were then analyzed with analysis of covariance (ANCOVA) in Minitab, with fish fork length (FL) and mass as covariates. Both linear and exponen.tial regression equations were fitted by least squares to the versus speed data for each individual. A repeated-measures multiple regression analysis in SAS (version 6..12) was used to determine whether the relationship between and speed differed significantly between the two species when the effects of fork length and mass were accounted for. The initial statistical model tested for significant effects on of species, speed, fork length and mass and for all possible two- and three-way interaction terms: species×speed, species×fork length, species×mass, speed×fork length, speed×mass, mass×fork length, species×speed×fork length, species×speed×mass. Variables and interaction terms that did not contribute significantly to the multivariate regression were dropped, one at a time, from the model, and a new regression was calculated until a final.model that included only variables with significant effects on was obtained. In all statistical analyses, a significance level of P=0.05 was used.

The maximum sustainable swimming speeds at a gait characterized by continuous tail beats (U_max) for juvenile kawakawa Euthynnus affinis, ranging in size from 116 to 255 mm FL and from 24 to 265 g, were 45–110 cm s⁻¹ and 3.4–5.1 FL s⁻¹ (Table 1). No marked increase in U_max occurred at the hypothesized minimum size for endothermy of 207 mm FL (Fig. 2). Chub mackerel (Scomber japonicus), ranging in size from 140 to 240 mm FL and from 26 to 156 g, had maximum continuous swimming speeds of 70–120 cm s⁻¹ or 3.8–5.8 FL s⁻¹ (Table 1). In both species, U_max increased significantly with FL (Fig. 2) and with mass (Table 1), but there was no significant interspecific difference in the relationship between U_max and fish FL or mass (ANCOVA). The single yellowfin tuna (Thunnus albacares) (202 mm FL and 111 g) had a U_max of 80 cm s⁻¹ or 4.0 FL s⁻¹, similar to that of similar-sized kawakawa and chub mackerel (Fig. 2).

The relationships between (mg O min⁻¹) at 24 °C and swimming speed (cm s⁻¹) for the individual scombrids in the present study are shown in Fig. 3. These data were modeled with both linear and exponential regressions (Table 2). The theoretical mechanical cost of swimm.ing should increase exponentially with speed, with plots of versus swimming speed taking the form of a ‘J’ (Webb, 1998). However, empirical data for swimming fishes do not necessarily fit this model and, in most studies, metabolic costs increase with speed at a lower rate than is predicted (for a review., see Webb, 1998). In the present study, a linear increase in with speed was often found, with a linear regression modeling the data for approximately 60 % of the individuals as well as or better than the exponential regression (see r² values in Table 2). With a linear fit, COT_net is directly proportional to the slope of the linear regression (see Materials and methods). With an exponential fit, COT_net varies with speed, and it is possible to determine a minimum COT_net and the speed at which COT_net is minimal. To do so, one first subtracts the standard metabo.lic rate (SMR), which can be estimated by extrapolating the versus speed curve to zero speed (e.g. Dewar and Graham, 1994), from the at each speed. However, extrapolation of the data results in large confidence intervals around the estimates, and the extrapolated SMR values differ greatly depending on whether a linear or exponential curve fit is used (see Table 2). Therefore, for consistency, we chose to use the slope of the computer-generated linear regression of versus speed as an estimate of the average COT_net of each individual (Tables 1, 2).

In both species, as fish size increased, mass-specific net cost of transport (COT_net) decreased (Fig. 4) and non-mass-specific net cost of transport increased (Fig. 5). No marked change in COT_net was recorded at the hypothesized minimum size for endothermy of 207 mm FL (Fig. 4), and the relationships between COT_net at 24 °C and fish size did not differ significantly between kawakawa and chub mackerel (Figs 4, 5). The one yellowfin tuna studied had COT_net values comparable with those of similar-sized kawakawa and chub mackerel.

We also calculated the total cost of tra.nsport (COT_total) at each speed for each fish by dividing the (mg O₂ min⁻¹) at each speed by speed (cm s⁻¹) and then converting to J km⁻¹ kg⁻¹ using the conversion factors given in the Materials and methods section. The speed at which COT_total is a minimum varied considerably among individuals, ranging from 40 to 110 cm s⁻¹ (67±22 cm s⁻¹; mean ± S.D.) for the kawakawa and from 30 to 110 cm s⁻¹ (55±25 cm s⁻¹ mean ± S.D.) for the chub mackerel. There was no significant difference between these mean values (t-test).

Repeated-measures analysis of covariance was used to describe the interrelationships between at 24 °C (mg O₂ min⁻¹), speed (cm s⁻¹) and body size (both fish mass and FL) for the juvenile kawakawa and chub mackerel. In both species, at 24 °C increased significantly with both swimming speed and fish mass (P:s;0.0001); the rates of increase did not differ in. terspecifically. There was no significant effect of FL on when the. effects of fish mass and speed were accounted for. Total was significantly higher at a given speed and size in the kawakawa than in the chub mackerel, by an average of 1.19 mg O₂ min⁻¹ (P<0.0001). To illustrate these relationships, the minimum and maximum values for both fish mass and speed were u.sed to solve the final multiple regression for each species for ⁠, and these values were plotted as the four corners of planes in three-dimensional space, together with the data points for each individual fish (Fig. 6). The difference in position along the y-axis between the planes for the kawakawa and the chub mackerel represents the magnitude of the interspecific difference in total metabolic rate. Because the COT_net values do not differ significantly between the two species, the interspecific difference in total can be attributed to the standard metabolic rate being significantly greater in the kawakawa than in the chub mackerel. Thus, on average, SMR in the juvenile kawakawa exceeds that of the chub mackerel by 1.19 mg O₂ min⁻¹. Because of the limitations to estimating SMR by extrapolation to zero speed (see above), we did not calculate an SMR value for each individual. We consider the mean interspecific difference in the versus speed relationships (Fig. 6) to be the best estimate possible from these data of the difference in SMR between the juvenile kawakawa and chub mackerel.

The maximum red muscle temperature elevations (T_x values, Table 1) for the kawakawa ranged from 1.0 to 2.3 °C (1.5±0.3 °C, mean ± S.D., N=9). For the chub mackerel, red muscle temperatures were elevated by only 0.1–0.6 °C (0.4±0.2 °C, N=5).

This project is the first to measure directly the maximum sustainable swimming speed of tunas of any size under controlled conditions. Furthermore, it is also the first to compare the maximum sustainable speed and the cost of locomotion in tunas with those in a closely related, similar-sized ectothermic teleost at the same temperature. The results do not support the hypothesis that juvenile tunas are faster or more efficient swimmers than their ectothermic relatives. The kawakawa of the size range studied did not have higher maximum sustainable swimming speeds nor did they possess lower net cost of transport values than did the chub mackerel. In fact, the metabolic rate at a given speed and fish size and the standard metabolic rate were greater in the juvenile kawakawa than in the chub mackerel. The finding of similar U_max values in juvenile tunas and chub mackerel does support previous hypotheses (Bushnell and Brill, 1991; Block et al., 1992; Brill, 1996; Korsmeyer et al., 1996b) that the maximum sustainable swimming speeds of tunas should approximate those of other active teleosts and provides evidence against earlier proposals that tunas are capable of extraordinary sustained speeds (Walters, 1962; Yuen, 1966; Magnuson, 1978).

There are few published measurements that are directly comparable with the U_max data of the present study. Because swimming speed varies with both temperature (for a review, see Rome, 1995) and fish size, valid interspecific comparisons require that acclimation and measurement temperatures and fish size be closely matched. Furthermore, to elucidate the selective advantage of specific characteristics such as endothermy and thunniform locomotion, comparisons must be made with closely related taxa in a phylogenetic context. Thus, the present study directly compared the aerobic swimming performance of tunas with that of similar-sized individuals of a closely related ectothermic species under similar conditions. The U_max values for S. japonicus measured in the present study (Table 1) are similar to maximum sustainable speeds measured previously in S. japonicus and the Atlantic mackerel Scomber scombrus. Roberts and Graham (1979) swam S. japonicus (343–392 mm FL) at maximum sustainable speeds of 3.2–4.5 FL s⁻¹ (125–171 cm s⁻¹) in a 12 cm diameter swimming tunnel at 16.1–21.8 °C, but the duration that each fish swam at those speeds was not given. At 11.7 °C, S. scombrus (29–33 cm FL) maintained a maximum speed of 3.5 FL s⁻¹ for 200 min and 4.3 FL s⁻¹ for 30 min (He and Wardle, 1988). For kawakawa (38±1.4 cm FL, mean ± S.D.), individuals swam in a respirometer at a mean (± S.D.) velocity of 57±1.3 cm s⁻¹ (range approximately 40–70 cm s⁻¹) at 24±2 °C (Dewar and Graham, 1994). This is the only other study that we know of that attempted to measure maximum sustainable swimming speeds in tunas in a swimming tunnel.

A number of methods have been used to quantify the sustained or prolonged swimming performance of fishes (for reviews, see Beamish, 1978; Videler, 1993). Maximum sustainable swimming speeds are often measured using the Brett (1964) protocol to measure critical speed (U_crit) in which speed is incrementally increased until the fish fatigues, but such studies have not been conducted with scombrids. The U_crit calculation takes into account the fraction of time the fish swam at the final speed at which it fatigued. Although our procedures approximated this, the U_max values from the present study will be lower than those that use the Brett (1964) procedure to determine U_crit, because we did not push any fish to its fatigue point and we did not take into account the fraction of time that the fish swam at the speed at which it shifted to a burst- and-coast gait. Nevertheless, our U_max measurements are within the ranges of U_crit determined for other active species, primarily salmonids, of similar sizes (for reviews, see Beamish, 1978; Videler, 1993).

The U_max values for the scombrids in the present study are greater than voluntary speeds that have been recorded for the same or similar species. Larger kawakawa (approximately 42 cm FL) swam at speeds of 56–68 cm s⁻¹ in 7.3 m diameter tanks at 22.8–25.9 °C, but swam at 126 cm s⁻¹ when stimulated by food flavor added to the tank (Magnuson, 1970). Typical speeds for kawakawa (mean length 36 cm FL) swimming in a 7.2 m diameter tank at 23–26 °C were 2.0–2.4 FL s⁻¹ (Magnuson, 1973). Dizon et al. (1977) measured voluntary swimming speed versus water temperature, as water temperature was changed at a rate of 5 °C h⁻¹, in a doughnut-shaped tank with inner and outer diameters of 4.5 m and 6 m, respectively. At 23–25 °C, kawakawa (40–50 cm FL) swam at 1.25–1.70 FL s⁻¹. Euthynnus lineatus (46.0–54.8 cm FL) swam at mean speeds of 68–79 cm s⁻¹ (1.35–1.71 FL s⁻¹) in a 2.5 m long × 2.3 m wide × 2.2 m deep bait well (Schaefer, 1984). For mackerel, Schaefer (1986) measured voluntary swimming speeds of fish subjected to a 1 °C day⁻¹ temperature change in a 1.52 m diameter tank. When at 24 °C, chub mackerel (22–26 cm FL) swam at approximately 1.2 FL s⁻¹ (approximately 26–31 cm s⁻¹) at 13:00 h, prior to feeding, and at approximately 1.9–2.1 FL s⁻¹ (approximately 42–55 cm s⁻¹) at 17:00 h. Preferred speeds of schooling Atlantic mackerel (Scomber scombrus) swimming in a 10 m diameter gantry tank were 0.9–1.2 FL s⁻¹ and 0.9–3.5 FL s⁻¹ for two groups, 28.5–39.5 cm FL and 28.5–40 cm FL, respectively (He and Wardle, 1988). The highest speeds for the second group of S. scombrus were associated with feeding. Because all these measurements were made on fish in confined conditions, they may not represent typical voluntary speeds for scombrids in the wild.

The present study set out to test the hypothesis that tunas swim faster and more efficiently than their ectothermic relatives. The focus on juveniles was based on the hypothesized minimum size for endothermy of 207 mm FL in Euthynnus lineatus (Dickson, 1994) and on the need to conduct size-matched interspecific comparisons. However, the red muscle temperatures measured in the present study for E. affinis are lower than those recorded by Dickson (1994) for E. lineatus of similar sizes. This discrepancy is probably due to the difference in methods used to obtain the temperature data, but could also be due to different amounts of red muscle in the two species [11 % in E. lineatus 289–570 mm FL, N=14 (Graham et al., 1983) versus 8 % in E. affinis 360 mm FL, N=1 (Magnuson, 1973), for larger individuals]. The temperature measurements were collected with the same thermocouple thermometer for both species, but the E. lineatus data were collected after the fish had been chased at a moderate speed for approximately 1 min. The latter temperature measurements are similar to acute temperature measurements in the literature, which represent values from freshly caught fish (e.g. Carey et al., 1971).

The muscle temperature data for the kawakawa and mackerel in the present study were collected at the end of the experiment in a manner that results in T_x measurements that are probably more indicative of in vivo conditions. The mean red muscle T_x of only 1.5 °C in the juvenile kawakawa in the present study may not have been sufficient measurably to enhance aerobic swimming performance. On the basis of the temperature effects between 25 °C and 30 °C on contraction time for red muscle fibers isolated from yellowfin tuna (Altringham and Block, 1997), a red muscle temperature elevation of 1.5 °C in juvenile kawakawa would result in an increase in U_max of approximately 7 %. Given the variability in the data, it may not have been possible to detect an increase in speed of that magnitude in the present study. However, the kawakawa tended to have lower, not higher, U_max values than did the chub mackerel (Fig. 2). The ability of tunas to elevate muscle temperatures above water temperature increases with fish size. For example, Lindsey (1968) reported a T_x as high as 5.4 °C for a 485 mm FL kawakawa. Thus, we cannot rule out the possibility that, at larger sizes, when the kawakawa can elevate red muscle temperature to a greater degree, there will be a significant difference in maximum sustainable swimming speeds. Future studies of larger tunas and similar-sized, closely related ectotherms, such as the bonitos, swimming at the same water temperature, are needed to determine unequivocally whether endothermy enables tunas to increase swimming speed.

The present study compared swimming energetics of juvenile kawakawa and chub mackerel on the basis of versus speed versus fish size relationships described as multiple linear regressions. Because these variables covary, such an analysis is necessary to account simultaneously for the effects of speed and size. However, because few previous studies have used a similar type of analysis for fish of a similar size range and temperature, it is difficult to make compariso.ns with published data. Dewar and Graham (1994) measured as a function of speed and mass in larger kawakawa (366–400 mm FL, 770–1000 g) at a water temperature of 24.4±0.2 °C (mean ± S.E.M.). They provided us with their raw data so that we could compare the energetics data for the juvenile kawakawa with those for larger individuals in which muscle temperature should be elevated sign.ificantly above water temperature. The multiple regression of versus speed and mass for the kawakawa in the present study was solved for using the data of Dewar and Graham (1994), and the.se extrapolated values were compared with their measured values. Although we acknowledge the limits of this extrapolation, there was good agreement between the calculated and empirical data, with a mean difference of −0.14±0.44 mg O₂ min⁻¹. (mean ± s.e.m., 25 measurements). There are no published versus speed data that we know of for any Scomber species.

To estimate mean COT_net for each individual, we modeled the data with linear regressions, rather than a ‘J’-shaped exponential relationship (Webb, 1998). Many of the values fitted the linear model well, probably because we specifically chose to study a single gait, when fish swam with a steady, continuous tail beat, which resulted in speed ranges in the most linear portion of the J-shaped curve. We also started most experiments at a moderate speed, rather than at lower speeds, for the following reasons. Because tunas are obligate ram ventilators and are negatively buoyant, it was necessary to commence all respirometry trials at a speed high enough to ensure adequate ventilation and hydrodynamic equilibrium (Magnuson and Weininger, 1978). In in.itial trials, low speeds were used, and they resulted in high values. These low speeds often elicited erratic fluttering of the pectoral fins and unsteady swimming in the tunas (visual observations) which, in combination with the inefficiency of muscle contraction at low strain rates (Webb, 1998.), probably caused the elevated values. In those cases, measurements at the lowest swimming speeds were not included in the linear regressions used to estimate COT_net. All future trials commenced at a higher speed, because we were interested only in continuous, sustainable swimming (i.e. a single gait). Furthermore, mackerel actively ventilate their gills by buccal pumping at low speeds (Boutilier et al., 1983), whereas kawakawa do not.

Thus, the added cost of the .contractions associated with this behavior could increase the of the mackerels and thus bias the interspecific comparison at low speeds.

There was no significant difference in the incremental cost of swimming (COT_net) between the juvenile kawakawa and chub mackerel (Figs 4, 5). However, total metabolic costs and standard metabolic rates were higher in the kawakawa (Figs 3, 6). The high metabolic rates of tunas have been documented in the literature (Gooding et al., 1981; Brill, 1987; Brill and Bushnell, 1991; Dewar and Graham, 1994; for a review, see Brill, 1996), but it was not until the present study that tunas were compared with a closely related ectothermic teleost of the same size range at the same ambient temperature. Prior investigations have been limited in that they have had to compare their results with the only available data from similar-sized ectothermic fishes, which has been from salmonids. Salmonids are active ectothermic teleosts, but they differ morphologically from tunas and they are only distantly related to scombrids. When Dewar and Graham (1994) compared energetic data from yellowfin tuna (416±27 and 513±30 mm FL) at 24±2 °C with those of similar-sized salmonids at 15 °C (Brett, 1965; Bushnell et al., 1984), tunas had higher standard metabolic rates, but their net cost of swimming (COT_net) was actually lower than that of the salmonids. This provided support for the hypothesis that tunas were more efficient swimmers than ectotherms, possibly because of the streamlined bodies of the tunas, the thunniform mode of locomotion and/or warm aerobic locomotor muscle. However, the results from the present study, a more appropriate comparison because we have used a member of one of the tunas’ sister groups and studied both species under the same conditions, provides evidence that tunas are not faster or more efficient swimmers, at least as juveniles. Since the kawakawa swam with less lateral displacement than the chub mackerel, it appears that the thunniform mode of locomotion is not more energetically efficient than the carangiform mode used by mackerels (Donley and Dickson, 2000).

The finding of higher standard metabolic rates in the juvenile kawakawa compared with the chub mackerel is probably related to the high aerobic capacity of tunas (Gooding et al., 1981; Dewar and Graham, 1994; Dickson, 1996; Korsmeyer et al., 1996a; Mathieu-Costello et al., 1996; Brill, 1996). Tunas possess anatomical, biochemical and physiological characteristics that have led researchers to conclude that they are high-performance fishes (for reviews, see Dickson, 1995; Brill, 1996). The need to maintain the high-performance ‘machinery’ is probably, at least in part, responsible for the high standard metabolic rates of tunas compared with mackerels. Tunas have high gill surface areas with a thinner gill epithelium than most teleosts, including mackerels (Gray, 1954; Muir and Hughes, 1969; Hughes, 1984). This thin barrier enables oxygen transfer from the water to the blood to occur twice as fast in tunas as it does in other teleosts (for a review, see Bushnell and Jones, 1994), but may also result in high rates of ion flux across the gills, which should result in high osmoregulatory costs (Brill, 1996; Korsmeyer et al., 1996a). A recent study showed that the combined costs of running the Na⁺/K⁺-ATPase pumps in both the intestines and gills, which are primarily responsible for ion regulation in teleosts, account for 16 % of the total SMR in yellowfin tuna (J. Y. Swimmer and R. W. Brill, personal communication).

The heart of tunas is larger than that of other similar-sized ectothermic teleosts, including mackerels, and is capable of generating greater blood pressures (Brill and Bushnell, 1991; for reviews, see Bushnell and Jones, 1994; Farrell, 1996). A high cardiac output, coupled with the elevated hemoglobin and myoglobin levels found in tunas, provides for an enhanced oxygen delivery system (for reviews, see Bushnell and Jones, 1994; Dickson, 1996). The activity of the enzyme citrate synthase, which catalyzes the first step in the citric acid cycle and is an index of tissue mitochondrial density, was significantly higher in the red muscle of the juvenile kawakawa than in the chub mackerel used in this study (Herrick, 1999). Thus, maintenance costs for the cardiovascular system and aerobic locomotor muscle may be greater in the kawakawa than in the chub mackerel. In addition, the kawakawa in the present study were probably growing at a faster rate than were the mackerel because of the approximately threefold difference in feeding rates, which could have contributed to their elevated ⁠. It is also possible that the kawakawa experienced more stress during the experiments than the mackerel did, even though we attempted to minimize stress by handling the fish as little as possible and to control the amount of stress by handling both species in an identical manner. If this were true, it could have contributed to the higher metabolic rates of the kawakawa relative to the chub mackerel.

Although a recent study (Freund, 1999) found that routine rates of oxygen consumption at 20 °C of the eastern Pacific bonito (Sarda chiliensis) were similar to those of similar-sized skipjack tuna, swimming speeds were not cont.rolled and were not measured for the bonito. Future studies of versus speed versus fish size relationships of similar-sized bonitos at the same temperature are needed to determine whether members of tunas’ sister taxon have an elevated standard metabolic rate like the tunas or whether they are more like the mackerels. Only then will it be possible to ascertain whether high SMRs are associated with the tunas only, and if this trait is more likely to have evolved before or after the divergence of the tunas and bonitos. Such future studies will allow us to determine whether the high standard metabolic rates of tunas were a prerequisite for the evolution of endothermy.

The increased metabolic cost in tunas, without any measurable benefit in terms of swimming performance, at least in juveniles, leads one to ask, ‘why be a tuna?’ The answer to this question is not known, but tunas are one of the most abundant apex predators throughout the world’s oceans, despite the low quantity and patchy distribution of nutrients (Sund et al., 1981). Their sheer abundance and coexistence with other similarly derived piscivorous fishes suggest that this added metabolic cost could, at least in part, be overcome by other factors that contribute to their evolutionary success.

We thank J. M. Donley, A. C. Gibb and J. B. Graham for assistance with respirometry experiments, R. Brill, K. Cousins, R. Dollar and R. M. Laurs for support at the National Marine Fisheries Service Kewalo Basin laboratory in Hawaii, J. B. Graham and R. Shadwick for use of equipment in Hawaii, G. Hunt and D. Parsons for the design and construction of the respirometer, H. Dewar and J. B. Graham for providing data, K. Messer and J. M. Donley for help with statistical analysis and graphics, and J. Hoskinson for assistance in maintaining fish. A special thanks goes to Dale Simmons for his expertise in catching juvenile tunas in Hawaii, without which this work would not have been possible. Information provided by Dave Itano, Chuck Holloway and Kim Holland helped us to locate a source of juvenile tunas. The comments of M. H. Horn, J. B. Graham and two anonymous reviewers significantly improved the manuscript. Hexcel and Plascore, Inc. donated supplies used in the respirometer. Funding was provided by the CSUF Departmental Associations Council, the NIH Minority Biomedical Research Support and Minority Scientist Development Programs (no. S06GM08258 and no. R25GM56820) and the National Science Foundation (no. IBN-9318065).