A large swim tunnel respirometer was used to quantify the swimming energetics of the eastern Pacific bonito Sarda chiliensis (tribe Sardini) (45–50 cm fork length, FL) at speeds between 50 and 120 cm s-1 and at 18±2°C. The bonito rate of oxygen uptake(V̇O2)–speed function is U-shaped with a minimum V̇O2 at 60 cm s-1, an exponential increase in V̇O2 with increased speed, and an elevated increase in V̇O2 at 50 cm s-1 where bonito swimming is unstable. The onset of unstable swimming occurs at speeds predicted by calculation of the minimum speed for bonito hydrostatic equilibrium (1.2 FL s-1). The optimum swimming speed (Uopt) for the bonito at 18±2°C is approximately 70 cm s-1 (1.4 FL s-1) and the gross cost of transport at Uopt is 0.27 J N-1m-1. The mean standard metabolic rate (SMR), determined by extrapolating swimming V̇O2 to zero speed, is 107±22 mg O2 kg-1 h-1. Plasma lactate determinations at different phases of the experiment showed that capture and handling increased anaerobic metabolism, but plasma lactate concentration returned to pre-experiment levels over the course of the swimming tests. When adjustments are made for differences in temperature,bonito net swimming costs are similar to those of similar-sized yellowfin tuna Thunnus albacares (tribe Thunnini), but the bonito has a significantly lower SMR. Because bonitos are the sister group to tunas, this finding suggests that the elevated SMR of the tunas is an autapomorphic trait of the Thunnini.
Fishes in the family Scombridae such as the mackerels (tribe Scombrini),bonitos (Sardini) and tunas (Thunnini) exemplify structural design features for maximizing swimming performance(Lighthill, 1969; Webb, 1975; Magnuson, 1978; Nauen and Lauder, 2000; Donley and Dickson, 2000). Tunas, which are the most derived scombrids, possess two additional locomotor-enhancing functions. The first of these is the capacity for regional endothermy, which elevates muscle power(Carey and Teal, 1966; Carey et al., 1971; Graham, 1973; Johnston and Brill, 1984; Altringham and Block, 1997; Graham and Dickson, 2001). The second is the use of the thunniform swimming mode, which reduces drag by minimizing lateral body undulations while generating thrust with rapid oscillations of the high-aspect-ratio caudal fin(Fierstine and Walters, 1968; Lighthill, 1970; Webb, 1975; Lindsey, 1978; Altringham and Shadwick,2001).
The functional and structural bases for tuna endothermy and thunniform swimming reside in the tunas' myotomal architecture. Red myotomal muscle (RM,the slow-twitch aerobic fibers that power sustained swimming) occurs both more anterior in the body and closer to the vertebral column in tunas than in other fishes (Kishinouye, 1923; Fierstine and Walters, 1968; Graham et al., 1983). Tuna RM position has been hypothesized to have influenced thunniform locomotion through effects on body shape (Magnuson,1978; Graham and Dickson,2000) and on the mechanical linkage between myotomes and the caudal fin (Westneat et al.,1993; Shadwick et al.,1998; Ellerby et al.,2000; Graham and Dickson,2000). In addition, shifts in the pattern of RM vascular supply may have established the basis for counter-current heat transfer and for the origin of endothermy (Graham and Dickson, 2000, 2001).
Although a unique RM position and regional endothermy occur exclusively in the tunas, the sequence of evolutionary changes that took place from the less-derived mackerels and bonitos to the tunas remains unclear. Therefore,testing or distinguishing among hypotheses about the acquisition sequence of these features requires phylogenetically appropriate structural and functional comparisons with nontuna scombrids(Magnuson, 1973; Graham, 1975; Collette and Chao, 1975; Collette, 1978; Block et al., 1993; Graham and Dickson, 2000, 2001).
Several works have compared the swimming performance of mackerels to that of tunas. In contrast to the thunniform swimming mode of tunas, both chub(Scomber japonicus) and Atlantic (Scomber scombrus)mackerels use the carangiform swimming mode(Videler and Hess, 1984; Donley and Dickson, 2000). Although the chub mackerel is not endothermic, rapid swimming does transiently elevate both RM and white muscle temperatures(Roberts and Graham, 1979). Finally, and despite similar costs of locomotion, the standard metabolic rate(SMR, the minimum metabolic rate required for maintenance functions; Videler, 1993) of the chub mackerel is lower than that of the tunas(Sepulveda and Dickson, 2000; Shadwick and Steffensen,2000).
Although mackerel–tuna comparisons identify key differences between the two groups, most investigators regard the bonitos, which are the tuna sister group (Collette, 1978; Block et al., 1993), as more appropriate for evolutionary comparisons(Block and Finnerty, 1994; Altringham and Block, 1997; Ellerby et al., 2000; Graham and Dickson, 2000). Relative to mackerels, bonitos have many morphological specializations (i.e. body shape, gill surface area, growth rate and maximum size) that are more like those of tunas (Gray,1954; Campbell and Collins,1975; Collette,1978). Bonito RM is also in a more medial position than in the mackerels (Kishinouye, 1923; Graham et al., 1983; Ellerby et al., 2000; Graham and Dickson, 2000), and both bonito myotomal structure and the arrangement of the anterior and posterior oblique tendons (which transmit force from the myotomal musculature to the caudal propeller) are more similar to those of tunas(Ellerby et al., 2000; Westneat et al., 1993; Westneat and Wainwright, 2001;K. A. Dickson and J. B. Graham, unpublished). However, bonitos do not have the anterior RM position found in tunas and they are not endothermic(Carey et al., 1971; Graham, 1975; Graham et al., 1983; Altringham and Block, 1997;Graham and Dickson, 2000, 2001).
While the importance of additional studies with bonitos to increasing our understanding of the swimming adaptations of tunas has long been recognized(Godsil, 1954; Graham, 1975; Block and Finnerty, 1994; Altringham and Block, 1997; Ellerby et al., 2000; Graham and Dickson, 2001),access to live specimens by laboratories capable of conducting physiological studies with them has been rare (Altringham and Block, 1997; Ellerby et al., 2000). The availability of the eastern Pacific bonito Sarda chiliensis in southern California coastal waters in the summer of 2000 provided the opportunity to study the swimming energetics and kinematics of this species using the same water tunnel respirometer that had been used for the tuna energetics and kinematics studies reported by Dewar and Graham(1994a,b). This paper reports on bonito swimming energetics, and the companion paper(Dowis et al., 2003) describes bonito swimming kinematics.
Previous work on bonito energetics includes indirect V̇O2 estimates derived from mouth gape and swimming distance(Mendo and Pauly, 1988) and estimates of routine metabolic rate (RMR, which includes both spontaneous movements and locomotion; Beamish,1978) made for bonito swimming inside of a small annular respirometer, without speed control(Freund, 1999). The objectives of the present study were to quantify bonito swimming energetics and obtain swimming–V̇O2data over a range of controlled speeds. We also wanted to compare bonito swimming to that of its sister group, the tunas, by contrasting the relationship between V̇O2 and swimming speed in similar-sized fish to estimate the cost of transport and SMR. This approach has provided new insight into two important questions about scombrid physiology. (1) How do the costs of locomotion compare for tunas and bonitos?(2) Is the high SMR of the tunas a unique tuna trait or a synapomorphy of the tribes Sardini and Thunnini?
Materials and methods
Fish collection and maintenance
Eastern Pacific bonito Sarda chiliensis Cuvier 1832 (size range:30–40 cm fork length FL, mass 500–1000 g) were caught in waters off La Jolla, California, in July and August of 2000 and transported to the laboratory at Scripps Institution of Oceanography (SIO) in a 378 liter cylindrical tank containing oxygenated seawater. Fish were captured using barbless hooks and transferred untouched into the transport tank. In the laboratory, bonito were transferred, without touching, to an 8 m diameter, 1.2 m deep, aboveground, plastic-lined holding tank supplied with continuously flowing aerated seawater (18±2°C). Fish were held in this tank over the course of the study (90–150 days). They were conditioned to feed with live guppies (Scientific Hatcheries, Huntington Beach, CA, USA) and, once feeding, were fed daily to satiation (about 5–10% body mass) using a mixture of chopped fish and squid.
The variable-speed water-tunnel respirometer used in this study has been described previously (Graham et al.,1990; Dewar and Graham,1994a; Graham et al.,1994; Bernal et al.,2001). It consists of a 2.3 m×6.3 m (length×width)oval of 46 cm-diameter polyvinylchloride pipe with an in-series diffuser-contraction section and a 100 cm×51 cm×42 cm(length×width×height) working section. It has a total volume of 3000 liters and is powered by a 40 hp variable-speed electric motor with a fixed, low-pitch propeller. Initial flow analyses(Graham et al., 1990) using attached threads, observations of particle motion and dye streaming all confirmed a uniform speed and laminar flow field in the center and first half of the working section, which is where swimming fish spend most of their time(see below).
Care was taken not to touch the bonito during all steps involved in capture and transfer to the water tunnel. Prior to all experiments, food was withheld from the entire tank for 24 h. The experimental bonito was obtained by fishing with a baited, barbless hook. The hooked fish was immediately placed in the cylindrical transport tank containing oxygenated seawater and the manufacturer's recommended dose of Fritzyme®, a mucus coat protecting agent. We found that the presence of other fish in the transfer tank minimized the introduced bonito's swimming speed and the frequency of its collisions with the tank wall. Standard procedure therefore was to place between one and five `companion fish' (either chub mackerel Scomber japonicus, or topsmelt Atherinops affinis) in the tank prior to introducing the experimental bonito.
After 15 min the bonito and the water around it were scooped up in a soft plastic bucket liner and transferred into the working section of the respirometer. We found that the transferred bonito could better orient to the flow and had fewer erratic movements and bursts if a `companion fish' was also present in the working section at the time of introduction. Both chub mackerel and topsmelt, which swim well in water tunnels(Sepulveda and Dickson, 2000;C. Sepulveda, personal observation) were used as `working-section companions'. The companion fish was removed after about 60 min, once the bonito had begun to swim steadily. Covering the working section with a dark cloth and placing a light at its front end helped to keep the bonito swimming in the center of the channel. A mirror positioned at 45° over the working section facilitated remote observation of the fish for the duration of the study.
The bonito was then given a 4 h period of slow, steady swimming to allow its recovery from capture and tunnel placement stress. Previous scombrid respirometry studies have noted a marked change in fish behavior and V̇O2 after about 2 h in the working section (Dewar and Graham, 1994a; Sepulveda and Dickson, 2000). During recovery the bonito swam slowly (about 60 cm s-1) while fresh seawater flowed through the tunnel to maintain temperature and ambient O2 levels.
Following the recovery period, water inflow to the tunnel was stopped and the system was sealed. Measurements of bonito V̇O2 were then made at different swimming speeds by recording water-tunnel O2declination rate over a 30–60 min period of steady swimming. Tunnel water speeds were determined by calibration with a flow meter (model 2035,General Oceanics Inc., FL, USA). The respirometer-water O2 level was monitored with a polarographic O2 electrode (Yellow Springs Instrument, OH, USA) connected to a model 52 meter. For each bonito energetics study, ProComm data acquisition software was used to record both water speed and O2 concentration.
During a given experiment, the respirometer water temperature was maintained within ±1°C of the bonito holding tank, which was supplied with a continuous flow of fresh seawater and subject to subtle environmental fluctuations in temperature (i.e. solar heating, seasonal fluctuation). Collectively for all tests, the experimental temperature was 18±2°C. Respirometer water O2 concentration was maintained at or above 80% saturation. When fish respiration reduced chamber O2 levels to near the 80% saturation limit, the system was re-oxygenated by seawater flow and application of a gentle stream of bubbles from a compressed O2 cylinder.
V̇O2measurements were made on bonito swimming between 50 and 120 cm s-1. Tests of each fish began by swimming it at the lowest speed it could maintain (50–60 cm s-1) for 30–60 min while O2 declination was measured. Following this determination, speed was increased by 10 cm s-1 and V̇O2 again measured. This protocol was repeated until the bonito could no longer swim steadily or maintain position. Most studies terminated at this point. However,in two cases, replicate V̇O2 measurements were repeated at a lower speeds. Video records of the bonito swimming at the different test speeds were then made for a kinematics analysis(Dowis et al., 2003).
After each individual swimming test, the bonito was removed and the empty respirometer was resealed to measure the background bacterial respiration rate and instrumental drift. Throughout the background measurements, the respirometer motor was circulating the water at approximately 50 cm s-1 and the V̇O2 was recorded for approximately 3–6 h. The mean background V̇O2 value for all experiments was 62±2.3 mg O2 h-1 or 6–14%of the maximum measured oxygen consumption rate; the percentage was greatest at lowest speeds. All reported bonito V̇O2 data are background-corrected. After each experiment, the water tunnel was bleached and cleaned with fresh water.
Bonito plasma lactate levels were measured to provide an indication of the level of anaerobic metabolic stress associated with three different phases of the respirometry study: (1) pre-experiment, just after capture from the holding tank, (2) at the end of the 15 min period in the transfer tank, (3) at the end of all swimming tests. Phases 1 and 2 used bonito that, although not used in the respirometer, were handled in exactly the same way as were the respirometer fish. Bonito taken directly from the respirometer following the completion of all swimming studies were used for phase 3 measurements.
Blood samples were taken by quickly grasping the bonito behind the operculum, and holding it firmly while removing approximately 2 ml of blood via cardiac puncture with a 20-gauge needle attached to a 3 ml syringe. Both the needle and syringe had been flushed with heparinized saline,which also filled the dead space. Plasma obtained by immediately centrifuging the blood (5 min at 3000g) was stored at -80°C. Lactate assays were performed spectrophotometrically using a Sigma Diagnostics kit(Procedure 735). This method did not employ a deproteination step and may thus have indicated slightly higher lactate levels than actually occurred. The tests do nevertheless provide a relative index for the change in plasma lactate concentrations at the different experimental phases.
Swimming V̇O2data were obtained for 12 bonito ranging in body size from 0.98 to 1.49 kg(45-50 cm FL) (Fig. 1, Table 1). Swimming speeds ranged from 50 to 120 cm s-1, with all of the fish studied swimming at two or more test speeds, and eight fish swimming at four or more speeds. The best performing fish was bonito 10, which swam from 60 to 120 cm s-1. Replicate V̇O2 tests were performed for two individuals, bonitos 3 and 7, to validate the repeatability of the V̇O2measurements. The first V̇O2 measurement for bonito 3 was 275 mg O2 kg-1 h-1 at 60 cm s-1. After swimming at two higher speeds (70 and 90 cm s-1) for 2 h, the V̇O2 of this fish remeasured at 60 cm s-1 was 250 mg O2 kg-1h-1. The initial V̇O2 of bonito 7 at 50 cm s-1 was 259 mg O2 kg-1h-1; after swimming for 1 h at 60 cm s-1, its V̇O2 at 50 cm s-1 was 243 mg O2 kg-1 h-1.
Covariance analysis (ANCOVA; Minitab version 12) indicated a significant(P<0.05) positive relationship between swimming speed and V̇O2, but no effect of either body mass or FL on swimming V̇O2. Absence of a body size effect allowed pooling of data for individual bonito and facilitated comparisons with V̇O2 values obtained for tuna (Dewar and Graham,1994a).
Fig. 1 shows the expected exponential increase in V̇O2 with increasing speed for each bonito (Videler,1993; Webb, 1998),but also indicates a trend for bonito V̇O2 to level off or even increase at 50–60 cm s-1. In fact, five of the six bonito for which swimming V̇O2 data were obtained at 50 cm s-1 actually had a lower V̇O2 at 60 cm s-1. The tendency for bonito V̇O2 to remain higher than expected at slower speeds is evident in the combined V̇O2-speed relationship for all bonito at each test speed(Fig. 2). This curve is U-shaped; the mean V̇O2 at 50 cm s-1 is higher than the V̇O2 at 60 cm s-1, and the value at 60 cm s-1 is only slightly less than the rate at 70 cm s-1. Two additional features of bonito low-speed swimming are indicated in Table 1. First, nine of the 12 fish had an elevated V̇O2 at the lowest speed tested (either 50 or 60 cm s-1) than at the next higher test speed. Second, records for these same fish at the time of study all noted erratic swimming (i.e. bursting, lateral movements and rapid fluttering of the pectoral fins) at these lower speeds.
A higher than expected V̇O2 at low swimming speeds affects estimates of both cost of transport (determined from the instantaneous slope of the V̇O2–speed regression) and SMR (estimated by extrapolation of the regression function to zero speed) (Videler, 1993). It was therefore important to quantify the effect of an elevated V̇O2 at low speed by contrasting a semi-logarithmic regression equation for the complete bonito data set with a `corrected' regression (i.e. calculated after the selective removal of approximately one high-V̇O2,low-speed data point from each fish). The criteria used to justify the selective removal of a high-V̇O2,low-speed data point were: occurrence of a lower V̇O2 at the next highest test speed and the occurrence, in notes made at the time of study, of observations of erratic swimming (Table 1).
Fig. 3 shows the effect of selective data removal: the exponential form of the regression equation for the complete data set is y=107e0.015x. The corrected data set regression equation is y=70e0.020x. The corrected function has a higher correlation coefficient (r=0.84 versus0.74); however, there is considerable overlap of the 95% confidence intervals of the slope and y-intercept values of the two functions, indicating that they are not significantly different. These two functions do nevertheless demonstrate the marked effect of an elevated V̇O2 at low speed both in lowering the regression equation's slope (from 0.02 to 0.15, a 25%reduction) and increasing its y-intercept, by 53%. Furthermore,because the y-intercept value at zero speed is an estimate of SMR(Videler, 1993), this selective data removal reduces the calculated bonito SMR from 107 to 70 mg O2 kg-1 h-1.
Bonito cost of transport
Although Fig. 3 indicates differences between the complete and corrected bonito swimming speed–V̇O2regressions, a valid comparison of the bonito data with published tuna energetics data necessitated use of the complete bonito data set because no high-V̇O2,low-speed data points were removed in the tuna study.
To calculate gross cost of transport (GCOT), the mass-specific V̇O2 values were first converted to weight-specific values (by dividing by 9.8 m s-2, the acceleration due to gravity), and these were converted from mg O2 to Joules (J), using the oxycalorific coefficient of 14.1 J mg-1 O2(Tucker, 1975; Videler, 1993; Dewar and Graham, 1994a). Then, these values were divided by the speed at which the value was determined, to obtain GCOT in units of J N-1 m-1. Because J=N m, this expression of GCOT is dimensionless: (mg O2h-1 kg-1) (9.8 m s-2)-1 (14.1 J mg-1 O2) (cm s-1)-1 (1 h 3600 s-1) (100 cm m-1) (kg m s-2 N-1) =J N-1 m-1. The net cost of transport (COTnet)at each speed was calculated in the same manner, starting with the rate of oxygen consumption above SMR (mass-specific V̇O2 at each speed– SMR).
The graph showing bonito GCOT in relation to speed(Fig. 4A, note speed units are m s-1), has a U-shaped function. The function's minimum defines the optimal swimming speed (Uopt, i.e. the speed with the lowest energy cost per unit distance)(Videler and Nolet, 1990; Videler, 1993; Korsmeyer and Dewar, 2001). The Uopt of the bonito at both 18°C and 24°C is approximately 0.7 m s-1 (about 1.4 FL s-1). The corresponding GCOT at Uopt is 0.17 J N-1m-1 at 18°C and about 0.27 J N-1 m-1 at 24°C. Also shown in Fig. 4Aare regressions describing bonito COTnet at both 18°C and 24°C. These functions are nearly parallel(y=0.061e0.885x, r=0.99, at 18°C; y=0.0951e0.916x, r=0.99, at 24°C);the difference in the y-intercepts reflects the assumed increase in SMR at 24°C (see below).
Most tuna respirometry has been done at 24°C. Because we lacked the heating and temperature-control equipment required to acclimate bonito to 24°C, the 18°C bonito V̇O2 data were adjusted to 24°C, assuming a Q10 of 2.0 (Brill, 1979, 1987; Dewar and Graham, 1994a), for comparison with tuna data.
Fig. 4B compares the GCOT and COTnet values for the bonito and yellowfin tuna at 24°C(data from Dewar and Graham,1994a). At all speeds compared, the GCOT is higher in the yellowfin tuna than in the bonito. Due to its higher SMR(Korsmeyer and Dewar, 2001),the yellowfin also has a higher Uopt, approximately 1 m s-1 (2.3 FL s-1) versus 0.7 m s-1 (1.4 FL s-1) for the bonito. The GCOT at Uopt for the yellowfin tuna is greater than that of the bonito (0.46 J N-1 m-1versus 0.27 J N-1 m-1). The COTnet functions for the bonito(y=0.0951e0.916x, r=0.99) and yellowfin tuna(y=0.0375e2.07x, r=0.98) differ, with the yellowfin having both a greater slope and a lower intercept. However, because the 95% confidence intervals of these parameters overlap, the two functions are not significantly different.
The bonito standard metabolic rate (SMR), as determined by the complete data set (Fig. 3), does not differ significantly (i.e. there is overlap in the 95% confidence intervals)from that of similar sized teleosts (Fig. 5), the yellowtail (Seriola quinqueradiata) at 20°C(Yamamoto et al., 1981) or the salmon (Oncorhynchus nerka) at 20°C(Brett and Glass, 1973). However, when adjusted for ambient temperature differences, the bonito SMR(161±33 mg O2 kg-1 h-1) is significantly lower (P<0.05, t-test) than that of similar sized yellowfin, skipjack and kawakawa Euthynnus affinis tunas(Brill, 1979, 1987; Dewar and Graham, 1994a) and dolphinfish Coryphaena hippurus(Benetti et al., 1995).
Plasma lactate concentrations
Bonito plasma lactate levels were determined for three phases of this respirometry study: just after capture from the holding tank, after 15 min in the round tank with companion fish, and after the swimming tests. The lactate concentration measured in the post-transfer period (2.84±0.88 mmol l-1, N=4) was much greater than the pre-experiment(baseline-control) lactate concentration (0.11±0.03 mmol l-1, N=2) and significantly higher (two sample t-test, P<0.05) than the post-swimming experiment levels(0.26±0.08 mmol l-1, N=4).
This study quantifies the swimming energetics of the eastern Pacific bonito at nearly constant temperature and over a range of sustained speeds and provides the first comparative metabolic performance data for a species in the tuna sister group, tribe Sardini. We found that the bonito COTnetis similar to that of both tunas and other active teleosts, including another scombrid, the chub mackerel (Scomber japonicus, tribe Scombrini). The SMR of the bonito was also found to be similar to values in similar sized yellowtail and salmon, but was significantly lower than that of size-matched yellowfin, skipjack and kawakawa tunas and dolphinfish(Fig. 5).Although we recognize the need for bonito and tuna comparisons at the same experimental temperature,this study offers strong evidence that the elevated SMR of tunas, and not their locomotory costs, distinguishes the tunas from other scombrids and most other pelagic, high performance fishes.
Validation of the bonito V̇O2–speed relationship
Both capture and handling stress have the potential to increase fish metabolic intensity to the point of elevating muscle and plasma lactate levels, followed by a period of elevated aerobic metabolism required for the reconversion of lactate to glycogen(Milligan and Girard, 1993; Milligan, 1996). Validation of the bonito swimming V̇O2 data therefore required confirmation that handling and confinement in the respirometer working section did not result in an excessive build-up of lactate. Our findings that bonito plasma lactate levels were elevated during the handling phases of the experiments, but then fell to low levels during the swimming tests, indicate that the metabolic costs of lactate reconversion were not a significant component of the bonito swimming-V̇O2determinations.
Bonito V̇O2 was determined over speeds ranging from 50 to 120 cm s-1. Although the exponential rise in V̇O2 occurring with incremental speed increases above 60 cm s-1 is entirely consistent with theory (Videler,1993; Webb, 1998), Fig. 2 shows that the mean V̇O2 for bonito is minimal at 60 cm s-1 and elevated at 50 cm s-1. Because erratic fin fluttering and unsteady swimming were documented for slow-swimming bonito (Table 1), we attribute the increase in mean V̇O2 at speeds of 50 cm s-1 (and, to some extent, 60 cm s-1) to the metabolic costs associated with the additional movements required for low-speed stability. This interpretation is consistent with findings for other fishes at low speeds (He and Wardle,1986; Lucas et al.,1993; Webb, 1998; Freund, 1999; Sepulveda and Dickson, 2000),as well as for other animals (Withers,1992; Schmidt-Nielsen,1993).
Most scombrids, including bonitos, are negatively buoyant and must swim continuously, both for hydrodynamic lift(Magnuson, 1973) and ram ventilation (Roberts, 1975). Magnuson (1973) used comparative morphometric and body density data to estimate the minimum speeds required for bonitos and tunas to maintain hydrostatic equilibrium. Application of the Magnuson(1973) minimum speed equation to bonito (45–55 cm FL) in this study yields a minimum speed of approximately 1.2 FL s-1 or about 54–66 cm s-1, which corresponds to the range where unsteady motion and increased swimming costs occurred for the bonito in this study(Fig. 2).
Unstable swimming and COT and SMR estimates
Comparison of the complete V̇O2–speed data set to one modified by the selective removal of the ten high-V̇O2,low-speed measurements (Fig. 3)documents how unsteady swimming artificially increases the SMR estimate and lowers the estimated swimming cost (regression parameters of the V̇O2–speed relationship). In the case of the bonito, this data manipulation was warranted because of our direct observations of unsteady swimming at low speeds(Table 1) and the known correspondence between unsteady swimming and the bonito's minimum speed requirements for hydrostatic equilibrium (Magnuson, 1973, 1978). While this correction did not significantly change the parameters of either regression function(i.e. there is overlap in the 95% confidence intervals), it did improve the correlation coefficient for the V̇O2–speed function. Although the selected data set appeared to be more meaningful biologically, comparisons with data for other species, for which selective removal was not done, required that we operate with the complete data set.
Comparisons of V̇O2 in bonito
Two previous studies provide swimming V̇O2 estimates for the eastern Pacific bonito. Mendo and Pauly(1988) used the observational swimming speed data of Magnuson and Prescott(1966) to estimate the routine metabolic rate (RMR) [defined by Beamish(1978) as the combination of normal and spontaneous movements] for a 2.5 kg, 57 cm FL bonito swimming in a large aquarium at 22°C. By combining mean swimming speed(1.4 FL s-1) with the estimated volume of water traversing the gills (based on mouth aperture or gape), and assuming the extraction of 50% of the O2 in the water contacting the gills, these workers arrived at an estimated RMR of 160 000 cal day-1, or 800 mg O2 kg-1 h-1. This value is three times the V̇O2 that we measured for bonito swimming at 1.4 FL s-1 (260 mg O2 kg-1 h-1) and doubtlessly reflects the numerous assumptions underlying the Mendo and Pauly(1988) V̇O2 estimate.
Freund (1999) reported similar RMR values for bonito and skipjack tuna swimming in an annular respirometer at 20±1°C. However, speed was not controlled in that study and unsteady swimming was noted for both species. The mean RMR estimated for the bonito was 427 mg O2 kg-1 h-1, and a speed estimate for one bonito was 50 cm s-1(Freund, 1999). Although bonito size and experimental temperature were similar in that study to ours,it is difficult to compare the two. Inspection of Fig. 2 shows that 50 cm s-1 is too slow for stable swimming in bonito of the size studied,and this may account for the 1.4-fold higher V̇O2 measured by Freund (1999).
Bonito and tuna comparisons
Although much of our understanding of tuna morphology and physiology suggests that they should swim more efficiently than other fishes, no study,including this work with the eastern Pacific bonito, has successfully documented this difference (Sepulveda and Dickson, 2000; Korsmeyer and Dewar, 2001). This is paradoxical because tuna specializations,including a high degree of body streamlining, a unique RM position, tendon arrangement, thunniform swimming mode and endothermy, have all been hypothesized to augment aerobic swimming performance and presumably swimming efficiency (Carey and Teal,1966; Fierstine and Walters,1968; Graham,1975; Johnston and Brill,1984; Westneat et al.,1993; Altringham and Block,1997; Graham and Dickson, 2000, 2001; Donley and Dickson, 2000; Ellerby et al., 2000; Altringham and Shadwick, 2001; Dowis et al., 2003).
Studies of tuna swimming efficiency, however, have of necessity been carried out on juvenile or sub-adult fishes in swimming chambers(Gooding et al., 1981; Graham and Laurs, 1982; Graham et al., 1989; Dewar and Graham, 1994a; Freund, 1999; Sepulveda and Dickson, 2000; Donley and Dickson, 2000). In the case of water tunnels, it may be that the requirement to maintain position by matching swimming thrust production to water flow speed obscures the utility of structural and physiological specializations for swimming that, in a more natural setting, increase swimming efficiency. A free-swimming tuna,for example, has options for turning, gliding and changing depth, as well as responding behaviorally to physical features in the environment, such as density gradients and thermal fronts. In addition, schooling may augment swimming efficiency (Weihs,1973). Magnuson's finding(Magnuson, 1973) of allometric changes that lower the minimum speed requirements of larger tunas further suggests that increased size, in combination with swimming specializations,increases swimming efficiency in larger tunas.
Standard metabolic rates
Extrapolation of the Fig. 3regressions to zero speed yields SMR estimates for the bonito that are comparable to those of the salmon (Brett and Glass, 1973) and yellowtail(Yamamoto et al., 1981), but significantly lower than both similar-sized yellowfin and skipjack tunas swimming in the same respirometer system(Dewar and Graham, 1994a) and spinally blocked (paralyzed) yellowfin, skipjack and kawakawa tunas (Brill, 1979, 1987) and dolphinfish(Benetti et al., 1995)(Fig. 5). The dolphinfish is the only fish species shown to have an SMR comparable to that of the tunas(Benetti et al., 1995). Although it lacks an elevated tissue temperature, the dolphinfish is a highly active species with several specializations (i.e. rapid growth rate, high heart rate) related to its high-energy-demand, pelagic existence(Brill, 1996). Similarity between the bonito SMR and that of the chub mackerel(Shadwick and Steffensen,2000) suggests that, among the scombrids thus far studied, the tunas are unique in having a high SMR.
The basis for the tuna's elevated SMR has been the topic of considerable discussion (Brill, 1979, 1987, 1996; Gooding et al., 1981; Dewar and Graham, 1994a; Brill et al., 2001; Sepulveda and Dickson, 2000; Korsmeyer and Dewar, 2001; Graham and Dickson, 2001). First, an elevated SMR is consistent with higher maintenance costs associated with the tunas' relatively large gills and heart and with the elevated metabolic requirements of the tunas' warm muscle and other tissues. In addition, there may be functional costs requiring a high SMR, such as the osmoregulatory load imposed by a larger gill surface area (Brill, 1987, 1996; Benetti et al., 1995; Brill et al., 2001). Tunas also have the added costs associated with the aerobic requirements of a larger and thicker-walled ventricle that has both a high cardiac output and a high systolic pressure (Graham et al.,1983; Brill and Bushnell, 1991, 2001; Bushnell and Jones, 1994; Graham and Dickson, 2000). Added to these may be a higher aerobic cost associated with the more aerobic white muscle of tunas. As in other fishes, white muscle accounts for the greatest percentage of body mass, and tuna white muscle has a greater aerobic capacity than in other species (as indicated by the high activity of the aerobic enzyme, citrate synthase) (Dickson, 1995, 1996; Korsmeyer and Dewar, 2001). Finally, there are the additional metabolic costs for processes such as gastric evacuation, somatic and gonadal growth rates, and lactate processing,which all appear to be augmented in tunas relative to other scombrids and other pelagic teleosts (Magnuson,1969; Schaefer,1984; Perry et al.,1985; Weber et al.,1985; Hunter et al.,1986; Arthur et al.,1992; Bushnell and Jones,1994; Brill, 1996; Korsmeyer and Dewar, 2001). The high SMR of tunas therefore appears important in enabling them to sustain high energy production rates required by the combinations of endothermy and a high aerobic scope for activity (Priede,1985; Bushnell and Jones,1994; Brill, 1996; Brill et al., 2001; Bushnell and Brill, 2001; Korsmeyer and Dewar,2001).
The objectives of this study were to obtain swimming energetics data for the eastern Pacific bonito in the same respirometer that has been used to study tunas, and to compare bonito and tuna locomotor costs and SMR. Because the bonito is in the scombrid tribe Sardini, which is the sister group of the tunas (tribe Thunnini), comparisons of tuna–bonito swimming performance have the potential to delineate the pattern of character acquisition leading to tuna locomotor specializations, including endothermy. Although our swimming performance comparisons necessitated a temperature adjustment (from 18°C to 24°C) of the bonito swimming–V̇O2data, the resulting net cost of transport (COTnet) estimate for the bonito significantly overlaps with that of the yellowfin tuna and thus indicates comparable swimming costs over the range of speeds tested.
On the other hand, the finding of a lower SMR in both the bonito and the chub mackerel relative to tunas implies that, in tuna, elevated SMR is an apomorphy linked to this group's numerous structures and functions requiring high metabolic maintenance. It may be that studies on the more basal species of tunas, for example, the slender tuna, Allothunnus fallai, will elucidate the linkages between SMR, aerobic function and endothermy. Allothunnus has less RM than other tunas and some of the circulatory modifications required for endothermy, but whether or not it is endothermic is not presently known (Graham and Dickson,2000).
We thank the Director's Office, Scripps Institution of Oceanography, and the Birch Aquarium at Scripps for supporting C. Sepulveda and this project. Funds were also provided by NSF grants IBN-9607699, IBN-9973916 and IBN-0077502, the California State University (CSU) Fullerton Departmental Associations Council, and an intramural grant from the CSU State Special Fund for Research, Scholarship, and Creative Activity. Bonito were obtained under California Department of Fish and Game scientific collecting permits. All experimental protocols were approved by the University of California San Diego and California State University Fullerton Institutional Animal Care and Use Committees. We thank C. Chan, J. Donley, H. Dowis, and H. Lee for assistance with fish collection and maintenance and with the experiments. We also thank D. Bernal, F. Powell, J. Hunter, P. Hastings, R. Rosenblatt, and R. Shadwick,for reviewing drafts of the manuscript, and J. Videler and an anonymous reviewer for comments on the final draft.