Pectoral Fin Locomotion in the Striped Surfperch: I. Kinematic Effects of Swimming Speed and Body Size

The speed and size of an animal are important determinants of its locomotor behavior. For bony fishes, the influence of these variables on patterns of propulsor movement during swimming has been extensively documented (e.g. Bainbridge, 1958; Hunter and Zweifel, 1971; Webb et al. 1984). This work, however, has focused to a great extent on a single mode of swimming, axial undulatory propulsion (Webb and Blake, 1985), in which thrust is generated by waves of bending of the body and tail. Although it has been less well studied, swimming with the pectoral fins, termed labriform locomotion by Breder (1926), is widespread among bony fishes. Labriform swimming is the primary mode of locomotion for many substratum-associated taxa and is particularly common among the vast assemblage of reef-dwelling perciform fishes (Lindsey, 1978; Blake, 1983). Modulation of pectoral fin kinematics with increasing swimming speed has been described for adult perciforms which use labriform locomotion over a narrow range of low speeds prior to the initiation of axial undulation (Parsons and Sylvester, 1992; Gibb et al. 1994; Webb, 1994). Apart from the studies of Webb (1973) and Westneat (1996), the relationship between speed and swimming kinematics has not been examined in fishes that rely primarily on the pectoral fins for generating thrust. In addition, the influence of body size on the temporal and spatial patterns of fin movement by pectoral fin swimmers remains largely unexplored (cf. Archer and Johnston, 1989).

Among teleosts employing labriform locomotion, the surfperches (family Embiotocidae) utilize a specialized swimming mechanism involving the generation of positive and negative lift forces during each pectoral fin-beat cycle in a fashion analogous to that proposed for the wing-beat cycle of some birds in flight (Webb, 1973). Compared with other labriform swimmers studied, surfperches swim over an unusually wide range of speeds using pectoral fin propulsion alone (Drucker and Jensen, 1996). From birth, the large, well-developed young are capable labriform swimmers which come to occupy the same general spatial and trophic niche as adults (Holbrook and Schmitt, 1984; Schmitt and Holbrook, 1984). Reliance on the paired fins for propulsion over their entire range of swimming speeds, and throughout their entire life history, makes the surfperches ideal subjects for studying the effects of speed and scale on labriform swimming behavior.

As an extension of the study by Webb (1973) which describes the effects of swimming speed on patterns of pectoral fin movement in the adult shiner surfperch Cymatogaster aggregata, we examine below the relationship between body size and the speed-dependent kinematics of pectoral fin swimming in an ontogenetic series of striped surfperch Embiotoca lateralis. We focus on this developmental size series rather than making cross-taxonomic comparisons of differently sized adults in order to avoid the confounding effects of interspecific variation in body form, habitat and locomotor ability (Huey and Hertz, 1982; Nee et al. 1991; Promislow, 1991). Specifically, we address the following questions: (1) how do the gait parameters affecting swimming speed and thrust (in particular, fin-beat frequency and amplitude) change with speed and (2) how are these kinematic functions influenced by body size? The answers to these basic questions contribute to a general model for comparing kinematic and physiological variables among labriform swimmers of different size (Drucker and Jensen, 1996).

Striped surfperch (Embiotoca lateralis Agassiz) were collected by beach seine at Argyle Bay, San Juan Island, Washington, USA. Fish were housed at the Friday Harbor Laboratories of the University of Washington in 260 l and 2350 l flow-through seawater tanks at 12 °C. In order to avoid any effects of prolonged pregnancy on swimming performance in these viviparous fish (Dorn et al. 1979), we used only males for experimentation. Individuals bearing gill parasites or exhibiting frayed fins were excluded from the study group. The fish were fed shrimp and squid every other day for 2 weeks prior to swimming trials.

Surfperch swam individually in a large racetrack-shaped flume against a current generated by a motor-driven paddle wheel (Nowell et al. 1989). The working area (34 cm wide × 25 cm deep × 80 cm long) was situated at the end of an 8 m long straightaway on the side of the racetrack opposite the paddles. Over the relatively large distance between the paddles and the working area, turbulence in the current was substantially attenuated. To collimate the flow further, plastic baffles (1.3 cm square grid) were positioned perpendicular to the floor of the channel upstream from the working area. These design features resulted in nearly rectilinear horizontal and vertical velocity profiles in the area where the fish swam. At distances greater than 5 cm from the walls and floor of the working area, the flow velocity was within 5 % of the mean free-stream velocity. Velocity profiles and the relationship between flow velocity and paddle motor speed were determined by timing drops of Methylene Blue introduced into the working area. The water temperature was maintained at 12 °C by a continuous low-volume flow of fresh sea water through the flume.

Four size classes of fish were selected for study from the population maintained in the laboratory (N=3 per size class; mean standard length, SL ± S.D.: 5.7±0.3, 11.9±0.3, 16.9±0.2 and 23.2±0.2 cm). None of the individuals was trained to swim in the flume. Each fish was starved for 24 h prior to its swimming trial and then anesthetized lightly with tricaine methanesulfonate and its body length measured. To improve the precision of kinematic measurements, the translucent tips of the pectoral fins were marked with small spots (approximately 1 mm diameter) of an opaque solution of cyanoacrylate adhesive gel and chalk dust. Fish recovered in the working area of the flume and were acclimated for 30–60 min at a low speed (approximately 0.6 SL s⁻¹).

Swimming speed was increased in 0.2 SL s⁻¹ steps from 0.6 SL s⁻¹ to the speed at which the fish could no longer prevent itself from contacting the downstream baffle of the working area. The interval between velocity increments was 3 min. As has been reported for other species during swimming trials at increasing velocity (Bainbridge, 1962; Brett, 1964; Brett and Sutherland, 1965; Videler, 1981), striped surfperch exhibited spontaneous unsteady swimming behavior at initial low speeds. In order to allow accurate kinematic measurements at these speeds, the initial portion of the swimming trial was repeated after the subject had reached its maximum speed and had become accustomed to forced swimming in the flume. Visual and auditory cues were occasionally used to encourage steady swimming in the center of the working area.

The fish were illuminated by two 500 W tungsten lamps and video-taped simultaneously with two Sony video cameras (60 fields s⁻¹, shutter speed 1/2000 s) in lateral view through the clear acrylic wall of the working area and in dorsal view by means of a mirror mounted at 45 ° above the working area. A floating plate glass roof over the working area eliminated turbulence at the water surface and distortion of the subject’s dorsal aspect. Live video signals were sent to a Panasonic digital video mixer (WJ-AVE5) which synchronized and combined partial fields from both dorsal and lateral views. The resulting split-screen images were recorded through a time code generator to S-VHS format video tape for analysis.

From video-taped records of swimming trials, several kinematic variables were measured for each fish during steady locomotion. The pectoral fin-beat cycle was defined as beginning and ending with consecutive onsets of pectoral fin abduction. Each fin-beat period (T) consisted of a period of propulsive fin movements (T_pr), from the onset of abduction to the end of adduction, and a non-propulsive pause or ‘refractory’ period (T_r) between the end of adduction and the beginning of the following abduction (Fig. 1). Field-by-field analysis of dorsal video images allowed measurement of stride frequency (pectoral fin-beat frequency, f_p), which was taken as the number of pectoral fin beats performed during a selected analysis period divided by the total time elapsed in seconds. Similarly, caudal fin-beat frequency (f_c) was calculated from the number of tail beats (complete cycles of left and right lateral displacement) observed in the analysis period.

Because T includes a refractory period, pectoral fin-beat frequency calculated in this study and others (e.g. Archer and Johnston, 1989; Gibb et al. 1994) as T⁻¹ underestimates the rate of force development by the pectoral musculature. To provide a more accurate measure of the average rate of thrust generation, we also calculated a corrected pectoral fin-beat ‘frequency’ as T_pr⁻¹ (cf. Webb, 1973), a term analogous to the reciprocal of ground contact time for the feet of running vertebrates (Taylor, 1994). Uncorrected frequency, f_p, is reported in addition to to facilitate comparison with earlier studies.

Pectoral fin-beat amplitude (A) was digitized from lateral video images using an Apple Macintosh computer and the program Image, version 1.51 (National Institutes of Health, USA). Amplitude was measured as the excursion (in cm) of the pectoral fin tip perpendicular to the horizontal swimming path from the onset of abduction to the onset of adduction (Fig. 1). Amplitude could not reliably be determined for the smallest individuals and is reported below for fish of 12 cm and longer. The pectoral–caudal gait transition speed (U_p−c) was the highest swimming speed at which the fish could hold station in the current for 3 min by pectoral fin oscillation alone (i.e. without regular use of the caudal fin: f_c<0.3 s⁻¹).

The video sequences selected for analysis met the following criteria. At speeds up to the gait transition, the fish had to swim steadily (i.e. remain stationary relative to a fixed reference point) for three or more consecutive pectoral fin beats. At speeds above the gait transition, at which pectoral fin oscillation was supplemented by unsteady axial undulation, the fish had to match the current velocity (±5 %) for at least three consecutive bursts of tail beating. We further required that the pectoral and caudal fin tips remained at least 5 cm from the walls of the working area. At this distance, the gap:span ratio of the fins was 2.3 or greater, indicating that hydrodynamic wall effects had a negligible influence on swimming performance (Lighthill, 1979; Webb, 1993).

Striped surfperch exhibit two distinct locomotor patterns during swimming trials at increasing velocity. These patterns are qualitatively similar to those described by Webb (1973) for the shiner surfperch Cymatogaster aggregata. At low to intermediate swimming speeds, pure labriform locomotion is employed in which the pectoral fins move synchronously through repeated cycles of abduction and adduction to generate forward thrust (Fig. 1). The trunk and tail do not undergo undulatory bending. At higher speeds, the pectoral–caudal gait transition occurs: undulation of the posterior trunk and caudal fin begins to supplement oscillation of the pectoral fins (see also Archer and Johnston, 1989; Gibb et al. 1994; Webb, 1994). Although the pectoral and caudal fins are occasionally used at the same time above U_p−c, the typical pattern is one of alternating pectoral and caudal fin oscillation. This swimming mode resembles the ‘burst-and-coast’ swimming behavior exhibited by axial undulators (Weihs, 1974; Videler and Weihs, 1982). Each intermittent burst of tail beating propels the fish forward and allows a period of coasting during which the body remains straight and the pectoral fins are fully adducted. Prior to the next burst phase, as the fish drifts backwards in the current, the pectoral fins oscillate rapidly. As opposed to most other labriform swimmers studied, which utilize body undulation alone at the highest swimming speeds (Archer and Johnston, 1989; Webb, 1994), surfperch employ pectoral fin oscillation, at least intermittently, over their entire range of swimming speeds.

For all sizes of Embiotoca lateralis studied, the relationship between uncorrected pectoral fin-beat frequency and swimming speed is non-linear (Fig. 2). Stride frequency increases slowly at low labriform swimming speeds and more rapidly at intermediate speeds, attaining a maximum at or soon after the pectoral–caudal gait transition. In contrast to f_p, the corrected pectoral fin-beat frequency increases linearly with swimming speed up to U_p−c. As caudal fin-beat frequency increases and axial undulation becomes the primary means of generating thrust, both measures of pectoral fin-beat frequency reach a plateau or decrease slightly.

Pectoral fin-beat amplitude expressed as a proportion of standard body length (A/SL) increases up to approximately 60 % U_p−c and then reaches a plateau, prior to the gait transition (Fig. 3). Unlike frequency and amplitude, the proportion of each pectoral fin stroke period occupied by propulsive fin movements (T_pr/T) decreases over low swimming speeds, falling from 0.78±0.05 to 0.45±0.04 on average (± S.D.) across the four size classes. This minimum coincides with the plateauing of fin-beat amplitude. Plots of T_pr/T versus speed show a second inflection point at U_p−c, at which speed a local maximum is attained (Fig. 3).

The influence of body size on the frequency of pectoral fin movement during labriform swimming is illustrated in Fig. 4. To maintain a given absolute swimming speed, smaller fish generally require higher f_p than larger fish (Fig. 4A). Similarly, the corrected pectoral fin-beat frequency is inversely related to body size. The rate of increase in with absolute speed decreases with size, but for all sizes is linear up to U_p−c (Fig. 4B). When speed is expressed in terms of standard body lengths traveled per second, however, these trends are altered: large fish (17 and 23 cm) exhibit a higher f_p than do smaller fish (12 cm) at speeds approaching U_p−c (Fig. 4C). As body size increases, the range of speeds (SL s⁻¹) over which the pectoral fins are used for propulsion is reduced. This gait compression (cf. Webb, 1994) is accompanied by an increase in the rate of change in f_p with swimming speed. Consequently, the curves describing the relationship between f_p and length-specific speed for the largest fish studied tend to cross those for smaller fish (Fig. 4C).

Unlike fin-beat frequency, length-specific amplitude does not vary substantially with body size at a given absolute or relative swimming speed. In fish varying twofold in length, A/SL increases at a roughly size-independent rate and reaches a plateau at 0.35–0.40 (Fig. 5).

The timing and magnitude of propulsor excursions in surfperch change with swimming speed in a manner generally similar to that reported for fishes employing axial undulation. For species relying on caudal fin propulsion, it has been demonstrated that length-specific fin tip amplitude is nearly speed-independent (Webb, 1975; Videler, 1993) whereas stride frequency increases linearly over a fish’s entire range of swimming speeds (Bainbridge, 1958; Hunter and Zweifel, 1971; Webb et al. 1984). For Embiotoca lateralis swimming in the labriform mode, increases in speed are also achieved by a steady increase in stride frequency (Fig. 2) while amplitude remains a roughly constant proportion of body length at all but the lowest speeds (Fig. 5).

A notable kinematic difference between the two swimming styles exists in the speed-dependence of stride frequency. Unlike caudal fin swimmers, surfperches increase this gait parameter non-linearly (Fig. 4A; Webb, 1973). An important distinction, however, must be drawn between the stride frequencies traditionally measured for axial and appendicular swimmers. The pectoral fin-beat period T includes a non-propulsive refractory period of variable duration, and thus f_p (T⁻¹) reflects the time-averaged frequency of fin oscillation, but not the rate at which forward thrust is generated. The tail-beat period of axial swimmers, by contrast, does not ordinarily contain a kinematic refractory period during steady locomotion (e.g. Bainbridge, 1963) and thus is more directly comparable with T_pr than with T. In contrast to f_p, the corrected pectoral fin-beat frequency (T_pr⁻¹) shows a linear increase with speed up to the gait transition (Fig. 4B), similar to the pattern reported for tail-beat frequency of caudal fin swimmers. This example illustrates that analysis of the timing of the propulsive phase of the stride cycle is required to select appropriate measures of frequency for interspecific comparisons.

Embiotoca lateralis increases swimming speed by modulating gait parameters in two distinct phases. Despite a decrease in the proportion of each pectoral fin-beat period occupied by propulsive fin movements, fin-beat amplitude and rise over speeds up to approximately 60 % U_p−c (Figs 2, 3). As amplitude reaches its maximum, continues to increase linearly (Fig. 2) and T_pr/T rises (Fig. 3), presumably to augment thrust at speeds nearing U_p−c. The plateauing of A/SL, followed by at higher speeds, and the peaking of T_pr/T at U_p−c may signify that E. lateralis can no longer effectively modulate pectoral fin kinematics to increase swimming speed and must instead switch to axial undulatory propulsion.

The physiological limit reached at the gait transition speed is probably related to the optimal velocity of shortening of the pectoral musculature. In fishes employing slow axial undulation, the shortening velocity (V) of aerobic myotomal muscle fibers increases with swimming speed over a narrow range in which mechanical power output is maximized (20–40 %V_max, the maximum contraction velocity: Rome et al. 1990). At higher swimming speeds (i.e. when V exceeds the upper limit of the optimal velocity range), the power generated by red muscle decreases substantially and a change in gait occurs involving the recruitment of white myotomal muscle to supplement the power output of the red fibers (Rome et al. 1988, 1990). Thus, axial undulators switch gaits when the major aerobic muscle fiber type being used can no longer generate power sufficient to sustain locomotion.

We hypothesize that the pectoral–caudal gait transition of labriform swimmers is driven by a similar mechanism. The product reflects the shortening velocity of pectoral muscle fibers (cf. Webb, 1973) and increases linearly with swimming speed (Fig. 6), as has been demonstrated for V of the red myotomal fibers of axial undulators (Rome et al. 1990, 1992). In contrast to myotomal fibers, for which V increases by a factor of two within a gait (from 0.2V_max to 0.4V_max), surfperch pectoral muscle undergoes a three-to fourfold increase in (Fig. 6), indicating that fibers of increasing speed may be recruited as swimming speed increases. At U_p−c, the fastest aerobic fibers used probably exceed the value of V for peak power and reach a limiting velocity at which power begins to decline. To swim faster without a further decrease of mechanical power output, labriform swimmers recruit anaerobic myotomal muscle and initiate tail beating.

The influence of body size on the linear rate of increase in the frequency of propulsive pectoral fin movements (Fig. 4B) matches that observed for tail-beat frequency in axial undulators (Hunter and Zweifel, 1971; Webb et al. 1984). The curvilinear relationship between the time-averaged frequency of pectoral fin oscillation and absolute swimming speed shows a similar general size-dependence (Fig. 4A). This pattern differs, however, when frequency is plotted against speed expressed in terms of body lengths per second, a common practice in studies of fish locomotion (e.g. Bainbridge, 1958). In particular, large surfperch exhibit higher f_p values than do smaller ones at high speeds (Fig. 4C).

The central messages of the present paper are that kinematic comparisons across body size yield different information according to the speed at which the comparison is made and that, furthermore, not all measures of speed allow valid comparisons. At an absolute speed at which large fish swim slowly, small fish approach the pectoral–caudal gait transition (Fig. 4A,B: 0.15–0.20 m s⁻¹). Even when speed is corrected for body length, fish of different sizes may not be at comparable levels of activity. A speed of 1.8 SL s⁻¹, for example, represents for small striped surfperch (6 cm SL) an intermediate labriform swimming speed at which f_p increases relatively slowly, but for larger fish (23 cm) is U_p−c at which the fastest rate of increase in f_p occurs (Fig. 4C). At such a speed, the stride frequencies of fish of different sizes cannot be meaningfully compared.

How then may the impact of body size on swimming kinematics be assessed? At the pectoral–caudal gait transition speed, labriform swimmers of different sizes are assumed to be at equivalent levels of exercise (Drucker and Jensen, 1996). Kinematic comparisons across size may accordingly be made at U_p−c or any percentage thereof (cf. Brett, 1965; Webb, 1971). The effect of body size on gait parameters differs markedly when speed is expressed as a percentage of U_p−c as opposed to SL s⁻¹. As an example, the product shows a distinct, but misleading, dependence on body length when plotted against length-specific swimming speed (Fig. 6A). The term increases in a size-independent fashion when plotted against speed in terms of percentage U_p−c (Fig. 6B). Assuming that the relevant muscle lever arms remain in proportion to body length, should reflect muscle shortening velocity in labriform swimmers of different sizes (Webb, 1973). The square of the term is also expected to be related to the thrust generated by the propulsive wave of pectoral fins of different sizes (Webb, 1971; Wu, 1977). Thus, the method used to express swimming speed influences both the conclusions that may be drawn about the effects of size on kinematic trends and the predictions that may be made about the size-dependence of the underlying physiological and mechanical patterns.

The following individuals are gratefully acknowledged for their critical commentary on the manuscript: F. A. Jenkins, Jr, J. H. Jones, K. F. Liem, R. L. Marsh, T. J. Roberts and C. R. Taylor. L. Self provided indispensable technical assistance. Access to the flume was kindly afforded by P. Jumars and D. Willows. Research support was furnished by the Lerner-Gray Fund for Marine Research, the Putnam Fund of Harvard University and NSF BSR-8818014 to K. F. Liem.