Studies of Tropical Tuna Swimming Performance in a Large Water Tunnel: III. Kinematics

The wave characteristics described above classify the broad spectrum of fish swimming styles and indicate the relative importance of the body segments and the paired and medial fins in thrust generation (Webb, 1975; summarised by Lindsey, 1978). In the anguilliform mode, for example, the propulsive wavelength is short and the amplitude along the entire body relatively large, which indicates that the body segments are primarily responsible for thrust generation (Gray, 1933). At the other extreme, the thunniform swimming mode employed by tunas has a long propulsive wavelength and the majority of lateral motion occurs caudally. The high-aspect-ratio lunate tail is the primary propulsor and anterior lateral movements (e.g. yaw) contribute only to drag.

Although thunniform swimming has been categorized in general terms (Fierstine and Walters, 1968; Webb, 1975; Magnuson, 1978), obtaining a more complete understanding of tuna swimming mechanics has been complicated by the limited opportunities for close examination of stably swimming fish. Nevertheless, unique muscular and biomechanical characteristics have been invoked to explain the means by which tuna attain high burst velocities (Brill and Dizon, 1979; Wardle and Videler, 1980). Related questions concerning thrust generation and swimming performance at sustained speeds have been complicated by the unavailability of a detailed description of tuna swimming kinematics. The previous inability to control U has also limited studies.

Much of the current information about the kinematic variables associated with the two-wave system discussed above (beyond the relatively simple measurements of TBF) is derived from a study by Fierstine and Walters (1968). These investigators examined one tail-beat cycle at five speeds for kawakawa (Euthynnus affinis) swimming in large circular tanks. Other important studies by Magnuson have examined the components of drag and thrust, the elements of tuna caudal movements, the actions of the finlets, the role of the pectoral fins in developing lift and, concurrently, the minimum swimming speed (U_min) required to maintain hydrostatic equilibrium (Magnuson, 1970, 1973, 1978). The most complete summary of the data available on scombrid swimming mechanics is provided in the paper of Magnuson (1978).

This paper reports observations of tuna swimming kinematics carried out on fish swimming stably at controlled U and has three primary goals. First, to establish the best estimates for the basic kinematic variables for tunas and to compare these with values determined for other teleosts. Second, to examine the functional significance of the anatomical and morphological adaptations of tunas for swimming. Third, to estimate the maximum swimming speed of tunas on the basis of swimming kinematics and muscle mechanics.

Kinematic observations were made on yellowfin tuna (Thunnus albacares) swimming at controlled U and ambient temperatures (T_a=24–28°C) in a large (3000 l) recirculating water tunnel (Graham et al. 1990). Details about the water tunnel and the methods for handling tunas are reported in Dewar and Graham (1994).

For fish that were swimming steadily at a constant U, TBF was determined using a stopwatch to record the time required for 20 tail-beats. In general, five replicate determinations were made 3–4 times at each experimental U. l was then calculated from the relationship between TBF and U. Because this method is relatively simple, the sample size for TBF and l is greater than for the other variables. Also, data were recorded for a larger number of fish.

Video recordings of stably swimming fish were digitized to quantify the caudal amplitude, yaw, C, the propulsive wavelength and the pectoral-fin sweepback angle. A mirror mounted at a 45° angle over the transparent top of the water tunnel’s working section made it possible to record the fish’s dorsal profile. Video recordings were made over a 2–5 min period at each experimental U using a Sony FX-800 video 8 camera. Shutter speeds up to 1/200 s were used, depending on U and the available lighting. While it would have been ideal always to record the movements of the entire body, this was often precluded by the requirement not to disturb the fish by uncovering the upstream region of the working section.

Selection of video segments for digital analysis (2–5 min videos) was based on several criteria. Sections were chosen only when the fish was away from both channel walls and its position, both parallel and perpendicular to flow, remained constant. Also, only segments in which the view was from precisely overhead were analyzed. The chosen segments spanned a minimum of 2 s and at least five tail-beats.

Prior to digitization, selected video 8 segments were transferred onto VHS format and a time code incorporated to label each frame numerically. A Panasonic (model no. AG-1960) and a JVC (model no. HR-S26700U) VCR enabled advancement of the video images frame by frame (VHS, 30 frames s⁻¹) or field by field (Panasonic, 60 fields s⁻¹, used when U>100 cm s⁻¹). A frame grabber transferred the selected video images to the computer monitor, where the Cartesian coordinates for specified parts of the fish’s body were determined. The same points on the fish were digitized in successive images to permit examination of their temporal progression.

Up to five points on the tuna’s body were digitized in each video image, depending on the extent of the body visible. Ideally, the tip of the snout, the base of the pectoral fins, a point close to the second dorsal fin, and both the proximal base and distal tips of the caudal fin were digitized. When visible, the second dorsal fin served as a natural marker; otherwise, the mid-point of the body positioned over a perpendicular grid line was used. For a 44 cm yellowfin swimming at 46 cm s⁻¹, the outline of the entire body was digitized through one propulsive cycle.

Fig. 1 shows all TBF data collected for the three size classes of yellowfin tuna as a function of U. Data for the size classes [separated on the basis of fork length (L)], together with the least-squares regression lines, are given in Table 1. Regression analyses indicate a significant increase in TBF with U. Examination of the differences in slope suggests an inverse relationship with L. The slope for the largest group (53 cm L) is significantly less than for the 32 cm and 42 cm groups, and the elevations between the two smaller groups differ significantly (P<0.05, ANCOVA).

Regression analysis also demonstrates that l is directly dependent on U for all three groups (Fig. 2). (It should be noted that, although the data are examined using linear regression analysis, at higher U the l will probably reach a maximum as it approaches the value for the propulsive wavelength.) Comparison of the three size classes of yellowfin shows an inverse relationship between slope and L. Also, the percentage of the data range occurring below U_min is greatest for the smaller yellowfin. [Note that l and the other kinematic variables that change with size are reported as length-specific values. For example, if the distance travelled per tail beat is 30 cm for a 50 cm tuna, l is 0.6L⁻¹ (30 cm/50 cm).]

Fig. 3A shows the dorsal view of the lateral undulations of a 44 cm yellowfin tuna at 1/15 s intervals through one tail-beat cycle (U=46 cm s⁻¹). The same images are superimposed in Fig. 3B. Fig. 3C shows the individual waves obtained by following the lateral excursion of the four points on the body (Fig. 3A) through two cycles. This formed the basis for determination of all kinematic variables. Digital analyses conducted on all fish are summarised in Table 2, which shows the mean L, the number of individual fish and video segments examined, as well as the mean and range of U used for each yellowfin size group.

From the wave patterns shown in Fig. 3C it is possible to determine the maximum excursion of points 1–4. (The base of the pectoral fins is not included because without a marker it was not possible to select a consistent point in successive frames.) The caudal amplitude is the distance between the lateral extremes of wave no. 4. Least-squares regression analyses for each size class and on the combined data for all fish indicate no effect of U on amplitude, which ranges between 0.17 and 0.20L¹(Table 3). The apparent decrease in amplitude with body size is not significant (P<0.05; non-parametric rank sum test).

Quantification of C also permits estimation of the propulsive wavelength (λ) (λ=CXTBF⁻¹). The propulsive wavelength is independent of U and the mean values (Table 4) range from 1.23 to 1.29L⁻¹. The slight differences between the three groups are not significant (P<0.05, non-parametric rank sum test).

To avoid potential complications imposed by the pivoting of the pectoral fins (apparent in Fig. 3B), the sweepback angle was measured (Fig. 4) at the lateral extremes and at the mid-point of the caudal propulsive cycle. The means of these three values, determined for all yellowfin, are plotted as a function of U in Fig. 4. Least-squares regression analysis on all data indicates that the sweepback angle increases significantly with U [sweepback angle=24+0.28U (cm s⁻¹) (N=30, r²=0.69)].

The movements of a swimming fish represent the cumulative interactions of its locomotor infrastructure and the physical forces imposed by the aquatic medium.

Quantification of movement can therefore provide insights into tuna swimming mechanics and the functional significance of the numerous morphological, muscular and skeletal characteristics that are considered to enhance swimming.

The localization and distribution of the locomotor muscle of tunas differs from that of other teleosts. Coincident with the strongly fusiform shape is the anterior localization of the tuna’s muscle mass and the insertion of the myotomes over a larger number of vertebrae (Nursall, 1956; Fierstine and Walters, 1968; Aleyev, 1977; Magnuson, 1978). The maximum cross-sectional area of the aerobic (red) locomotor muscle occurs at approximately 50% L, in comparison to the 75% L observed in many other fishes (Greer-Walker and Pull, 1975; Graham et al. 1983). Tuna red muscle also differs in that it lies adjacent to the backbone and not in a subcutaneous lateral wedge as in most other teleosts (Kishinouye, 1923; Carey and Teal, 1966; Graham et al. 1983). Although the importance of the central red muscle for endothermy is clear, the biomechanical implications of this arrangement remain to be determined. Lack of knowledge concerning the specific mechanical linkages between the propulsive musculature and the backbone, skin and caudal fin complicates this matter.

Another dominant component of force transmission and swimming mechanics is the compliance of the vertebral column. Rotation around intervertebral joints in tunas is limited by the articulation of zygapophyses (Nursall, 1956; Fierstine and Walters, 1968). Bending is thus limited primarily to the pre-and postpeduncular joints at the proximal and distal end of the caudal peduncle. The caudal peduncle itself is rigid owing to the bony lateral keels and the overlapping, recumbent neural and haemal spines (Nursall, 1956; Fierstine and Walters, 1968; Magnuson, 1978).

Fig. 3 allows examination of the undulatory movements of a yellowfin through a propulsive cycle. Although the successive profiles (Fig. 3A) give the impression of minimal anterior lateral motion, the superimposed images (Fig. 3B) illustrate considerable yaw. This becomes reduced at the base of the pectoral fins, but our analysis reveals that there is not a point of zero motion.

Although yawing occurs, most lateral motion takes place over the caudal region. Proceeding caudally from the base of the pectoral fins, the amplitude gradually increases and the envelope of lateral excursion does not extend beyond the point of maximal thickness until the caudal peduncle. Here, the dual joint system permits greater rotation, and a marked increase in amplitude occurs at the caudal tip. The pre-and postpeduncular joints and the stout ligaments are thought to focus the propulsive energy and motion generated by anterior musculature at the caudal fin (Nursall, 1956; Fierstine and Walters, 1968).

With the exception of the propulsive wavelength (and also C), our analysis reveals that all kinematic variables measured for the yellowfin are similar to those of other teleosts. A comparative discussion of each variable follows.

For tunas, as in most fishes, TBF increases with U (Bainbridge, 1958; Hunter and Zweifel, 1971; Webb, 1975; Webb et al. 1984). Although the range of TBF values is similar to that of other teleosts, our data show the slope for tunas to be less. Fig. 5 compares TBF data for the 53 cm yellowfin group with values determined by Hunter and Zweifel (1971) for 50 cm Pacific mackerel (Scomber japonicus), rainbow trout (Oncorhynchus mykiss) and jack mackerel (Trachurus symmetricus). The lower slope of the yellowfin tuna may stem from the considerable morphological adaptations for drag reduction or from adaptations affecting thrust production or propeller efficiency (discussed below).

The functional significance of TBF differences is further revealed by comparisons of l. Determination of l and its dependence on U is important for examining sustained swimming speeds and for estimating the relationship between maximum U and maximum TBF. In previous reports for tunas, the dependence of l on U has ranged from negative to positive (Yuen, 1966; Fierstine and Walters, 1968; Graham et al. 1989; Wardle et al. 1989). Our study, and that of Fierstine and Walters (1968), demonstrates that l increases with U (Fig. 2), which agrees with data collected for a number of other teleosts (Hunter and Zweifel, 1971; Webb et al. 1984). The range of l reported here is also similar to that of other teleosts [0.6–0.8 (Bainbridge, 1958; Webb, 1975; Webb et al. 1984)], but the capacity of scombrids to attain a l higher than 0.8L¹has been demonstrated (Wardle and He, 1988; Wardle et al. 1989). Our study has also revealed relatively low l values in smaller yellowfin (0.2–0.4; Fig. 2) at speeds below U_min. This may result from a decrease in swimming efficiency associated with the compensatory mechanisms employed to increase lift. Also, drag may increase at low U because the energy loss associated with inertial recoil is inversely related to TBF and consequently U (Lighthill, 1977; Webb, 1988). Unlike many other fish, tunas lack an alternative mode of propulsion at low U.

Although amplitudes determined in this study are less than the 0.21L¹value described by Hunter and Zweifel (1971) as typical for a number of marine teleosts, they fall within the range of values (0.13–0.21L¹) reported by other investigators (Bainbridge, 1958; Webb et al. 1984) and are similar to the value for giant bluefin (0.16L¹; Wardle et al. 1989). Reports of values up to 0.34L¹at a U of 8.2 L s⁻¹ (Fierstine and Walters, 1968), however, suggest that tuna are capable of greater amplitudes than measured here.

Fish generally achieve yaw reduction through an increased body depth, which resists the anterior lateral motion resulting from recoil forces (Lighthill, 1977; Magnuson, 1978; Webb, 1988, 1992). For tunas, it has been generally held that the structural and mechanical separation of the body from the caudal fin reduces recoil (Aleyev, 1977; Lighthill, 1975; Magnuson, 1978). This diminishes the need for an increased anterior depth and permits adoption of a fusiform body shape.

Contrary to the previous descriptions of tuna swimming mechanics, the yaw for yellowfin determined in this study is comparable to values reported for other teleosts. Our range of values (0.023–0.05L¹) is similar to the range 0.025–0.07L¹reported by Webb (1988, 1992) for the bluegill sunfish (Lepomis macrochirus), rainbow trout and tiger musky (Esox sp.). That the yaw for the yellowfin is not greater, given the lack of typical yaw-reducing adaptations, attests to the mechanical separation of the caudal fin and body, as well as to the contribution of the second dorsal and anal fins to the reduction of recoil (Magnuson, 1978). Our results also indicate that yaw, and thus the energy loss associated with recoil, decreases with U. This is expected because yaw is inversely proportional to both TBF and U×C⁻¹ (Lighthill, 1977; Webb, 1988, 1992).

The ratio of U to C is considered to indicate swimming efficiency (Lighthill, 1975; Wu and Yates, 1978; Webb et al. 1984). Comparison of yellowfin and other teleosts (Fig. 6), however, shows that the U×C⁻¹ ratio for the latter is approximately 40% higher, suggesting that tunas are less efficient swimmers. This is counterintuitive to expectations and to results from energetic studies indicating that tunas are more efficient swimmers (Gooding et al. 1981; Graham et al. 1989; Dewar and Graham, 1994).

The low U×C⁻¹ ratio for the yellowfin is attributable to its long propulsive wavelength, which results in a high C (C=wavelengthXTBF). Thus, even though TBF and l are comparable to those of other teleosts, the U×C⁻¹ ratio indicates that the yellowfin is a less efficient swimmer. This suggests that for comparisons made on fish with different propulsive wavelengths, U×C⁻¹ does not accurately portray efficiency. The U×C⁻¹ ratio does, however, indicate performance scope (i.e. U×C⁻¹ approaches 1 as U increases). Thus, the low values for tuna may only reflect the low range of test speeds used in this study in comparison with their potential swimming speeds.

As discussed above, propulsive wavelength is perhaps the most important kinematic variable. It defines swimming mode, indicates factors dominating thrust and drag, and sets the upper limit on l. Consequently, accurate assessment of this variable is essential. Previous investigators were forced to determine the propulsive wavelength from still images and the reported values range from 1 to 2L¹(Fierstine and Walters, 1968).

Use of digital analysis enabled more precise estimation of yellowfin propulsive wavelength. Our values (1.23–1.29L¹) exceed those of other sustained-swimming teleosts such as salmonids [0.7–0.9L¹(Bainbridge, 1958; Webb et al. 1984; Webb, 1988, 1992)]. The 30–60% longer propulsive wavelength for the yellowfin translates directly into a greater maximum l.

A longer propulsive wavelength distinguishes yellowfin (and probably all tuna) swimming from that of other teleosts and is attributable to a suite of morphological adaptations for locomotion. Both the rigid vertebral column and the insertion of myotomes over a greater number of vertebrae should reduce regional bending and thus lengthen the propulsive wavelength. As a result, muscular force is focused on the caudal fin and not on body segments which are ineffective for thrust generation. Also, assuming that the greatest muscular cross-sectional area is primarily responsible for the portion of the tail-beat cycle where the majority of thrust is generated, a specific phase difference between the time of muscle contraction and the sweep of the caudal fin across the mid-line is required. If the distance between the muscle and the caudal fin increases, then the propulsive wavelength must also increase in order to conserve the propulsive phase relationship. The external morphology associated with the distribution of muscle (i.e. fusiform body shape) should also increase swimming efficiency by reducing drag.

Although the locomotor advantages of a long propulsive wavelength are considerable, there are also disadvantages. For a fish with one or more waves on the body, the inertial recoil forces imposed by one part of the body wave going in one direction are countered by a portion of the body going in the opposite direction (Lighthill, 1977). This may help to explain the relatively high yaw observed for yellowfin at low U.

Although the sweepback-angle equation indicates that at U=235 cm s⁻¹ the pectoral fins will be completely adducted, the actual value of U is probably less; there is likely to be a point before the sweepback angle reaches 90° where the fins become ineffective for lift generation. In support of this, yellowfin swimming at 150 cm s⁻¹ sporadically adducted their pectoral fins. Complete pectoral-fin adduction may result in a 23% reduction of resistive forces; this is the estimated contribution of these fins to total drag (Magnuson, 1978). Whether energetic savings are incurred as the sweepback angle increases is difficult to determine because, although the reduction in surface area reduces friction drag, a higher U increases friction drag. Pressure and induced drag must also be considered.

The maximum U (U_max) of a fish is directly related to its greatest attainable l and TBF and can be calculated as U_max=l×2 T⁻¹, where T is the time (s) between nervous stimulation and the peak of muscular contraction for an unloaded muscle (Wardle, 1975; Brill and Dizon, 1979; Wardle and Videler, 1980). Examination of the components of this equation indicate that both a long l (discussed above) and a short T contribute to the tuna’s high burst swimming speeds. At equivalent temperatures, the T values for tunas and other teleosts are comparable (Brill and Dizon, 1979; Johnston and Brill, 1984); however, T is reduced in tunas as a consequence of endothermy (T is inversely related to temperature).

Even though the anaerobic (white) muscle of tunas does not contribute to heat production, thermal profiles indicate that the average white muscle thermal excess (T_x=muscle temperature minus T_a) can be half that of the red muscle (Carey, 1973; Graham and Dickson, 1981). Consequently, at high U, the red muscle T_x can reach 10°C and the white muscle T_x can reach 5°C (Stevens and Fry, 1971). Thus, the estimated U_max for a 40 cm tuna at a T_a of 25°C is 27 L s⁻¹ [1.2/0.044; T=0.022 s at 30°C (Brill and Dizon, 1979)]; this is within the range of maximum burst velocities reported for tunas (20–27 L s⁻¹; summarised by Magnuson, 1978). For comparison, the U_max for a similar-sized salmonid is 13 L s⁻¹ [0.8/0.06; T=0.03 s at 25°C (Brill and Dizon, 1979)]. Thus, both swimming mode and endothermy contribute to the high-burst swimming speeds of tunas.

We thank Drs J. R. Hunter, R. H. Rosenblatt, R. Shabetai, R. E. Shadwick and Chin Lai for commenting on drafts of this work. We also thank S. Zimmermann, P. Fields, S. Arce, C. Lowe, G. Spencer, D. Reisinger, K. Korsmeyer, R. Brill and the staff at the NMFS Kewalo Basin Research Facility for assistance. This research was supported by the US National Science Foundation (OCE-89-15927 and 91-03739). Additional funding was received from the William H. and Mattie Wattis Harris Foundation, the Academic Rewards for College Scientists Foundation, Inc., the Maurice Massarini Charitable Trust and the SIO Tuna Endowment Fund. We thank the Southwest Fisheries Science Center NMFS for facilitating our work at the Kewalo facility.