On The Function of the White Muscles in Teleosts at Intermediate Swimming Speeds

Fish myotomes contain two basic types of muscle fibre, red and white. Where these are separate the red fibres are found as a thin superficial layer below the skin along the mid-lateral aspect of the fish and form a thicker triangle of muscle at the peripheral extremity of the horizontal septum. Macroscopically these muscles are clearly distinguishable by their red colour. The differences between red and white muscles have been described and reviewed by Barets (1961), Boddeke, Slijper & van der Stelt (1959) and Bone (1964, 1966). Studies in a number of different disciplines have suggested a functional division between these muscle types at least similar to that found in the Anura (Kuffler & Vaughan Williams, 1953 a, b). Histochemical studies indicate that the red muscles metabolize fat anaerobically while in the white metabolism is by anaerobic glycolysis (Bone, 1966; George, 1962; Ogata, 1958 a, b). On this basis the white muscles have been implicated in vigorous short-duration activity. The much higher levels of oxidative enzymes (see also Ogata & Mori, 1964) and large stores of fat in red muscle point to these muscles being involved in sustained activity such as in the migration of salmon (Drummond & Black, 1960; George, 1962). Boddeke et al. (1959) separated fish according to their swimming behaviour into ‘sprinters’, ‘sneakers’, ‘crawlers’ and ‘stayers’ and compared the proportion and distribution of red muscle in representatives of each. They concluded that the white muscle fibres are suited to the behaviour of the ‘sprinter’, whilst the red fibres are suited for the sustained swimming of the ‘stayer’. ‘Sprinters’ are characterized by an almost total lack of red fibres.

Takeuchi (1959) described the existence of two distinct nerve-muscle systems in teleosts based on intracellular recordings. Barets (1961) obtained similar results from a range of teleosts and showed that the red muscle gave a plateau tension response related to stimulation frequency but failed to respond to a single shock, as in the frog slow-muscle system (Kuffler & Vaughan Williams, 1953b). In hagfish (Andersen, Jansen & Løyning, 1963) and carp (Hidaka & Toida, 1969) however, a slow twitch contraction was elicited from the red muscles by a simple junctional potential. Depolarizing agents applied to red and white muscles elicit different responses that support the distinction of these muscles into slow and fast respectively (Barets & Pecot-Dechavassine, 1959; Bone, 1966). Bone (1966) demonstrated by extracellular recordings that the white muscles of dogfish are involved in short-duration vigorous swimming movements that fatigue rapidly, while the red muscles are involved in slow sustained swimming. Rayner & Keenan (1967), who recorded electromyograms from skipjack tuna, concluded that there is a similar functional distinction between the red and white muscles in teleosts.

The white muscles of dogfish are focally innervated (Bone, 1964) and would be expected to give a fast singular response. In contrast the fast muscles of the majority of teleosts receive an extensive distributed innervation (Barets, 1961), conduct a propagated spike-potential that overshoots, and are considered to be polyneuronally innervated by 8−22 axons in the absence of a multiterminal innervation (Hudson, 1969). Differences in the K⁺ and Cl’ conductances between elasmobranch and teleost fast-muscle membranes have been described by Hagiwara & Takahashi (1967). Weak local contractions are elicited from fast fibres by junction potentials below threshold for the initiation of spike-potentials (Hudson, 1967). This has been confirmed in other teleosts, but is not found in elasmobranchs (Hagiwara & Takahashi, 1967). The twitch-muscle system of teleosts thus has properties in common with crustacean muscle fibres (Atwood, Hoyle & Smyth, 1965; Hoyle & Wiersma, 1958). A functional significance is suggested by the above features for the fast muscles of teleosts such that they may be involved in a variety of mechanical actions. In the present paper the possibility that the white muscles are involved at cruising speeds is examined by recording the activities of the red and white muscles of fish swimming at different speeds.

Rainbow trout, Salmo gairdneri Richardson, 26−34 cm long, were obtained from local fish hatcheries (Nailsworth, and Blagdon Power Station) and stored in exercise tanks supplied with a continuous flow of 8000 1 of re-circulated fresh water maintained at 15 °C. Fish were kept for upwards of 2 weeks before experiments to allow for acclimatization.

The slightly modified version of the Brett respirometer (see Brett, 1964) installed at Bristol University allowed the swimming movements of trout to be studied. A square-section fish chamber was fitted to facilitate observations (see fig. 1, Webb, 1971a). In this apparatus a constant volume of fresh water is propelled around a closed circuit at different speeds by means of a stainless steel centrifugal pump situated downstream from the fish. The details and flow characteristics of this respirometer have been described by Webb (1970). Grids at the divergence of the expansion cone break up turbulence so that a streamline flow of microturbulent water flows through the fish chamber. Water velocity was measured by a magnetic flow meter incorporated in the apparatus. Velocities measured by this method differed by less than 2% from those determined by a dual capacitative method, or by using a pitot-static tube (Webb, 1970). Water was oxygenated in an external tank after passing through a charcoal filter and returned to the respirometer via a constant-head apparatus. Water temperature was maintained at 15 °C + 1 °C by means of a 300 W radiant heater and a refrigerated cooling jacket.

The tail-beat frequency and amplitude of swimming fish were determined by means of a simple photoelectric movement monitor described by Hudson & Bussell (1972). The movement of a shadow cast by the fish’s caudal peduncle was translated into an electrical signal for display alongside electromyograms.

Diphasic extracellular recordings of muscle activity were made using four pairs of 40 s.w.g. Diamel-coated stainless steel wire hook electrodes. The final 1 mm of each hook was bared and the exposed tips were separated by 1·0−1·5 mm. The wires in each pair were glued together with Araldite for 10−20 mm behind the exposed tips to prevent rotation of the wires in the muscle and changes in configuration of the recording site. The other ends of the wires were bared and each was clamped in a tension spring to provide electrical continuity with the recording apparatus. Responses were recorded via a—c coupled differential amplifiers with a gain of 1000 and a 10 kHz filter, and displayed on an oscilloscope.

Fish were anaesthetized in a solution of MS 222 diluted 1 :2oooo. During operative procedures, conducted in air, the gills were perfused with an MS 222 solution of the same strength to maintain anaesthesia. Electrode pairs threaded into the barrels of hypodermic needles were implanted in the muscle by inserting the hypodermics through small cuts in the skin to the required position. When the hypodermics were withdrawn the electrodes hooked into the muscle and remained in place. Two pairs of electrodes were inserted on each side of the fish below the dorsal fin nearly equidistant from the snout - one was placed in the red muscle, the other in the deep epaxial mosaic (see p. 513) muscle. The eight insulated electrode wires were fed through a thin Polythene tube and sutured to the back of the fish immediately posterior to the dorsal fin to prevent entanglement and damage to the fish (Webb, 1970). Still anaesthetized, fish were introduced into the respirometer through a watertight port situated downstream from the fish chamber. The electrode wires were led out through an open tube set vertically in the cover of the watertight port, and connected to the tension springs. The apparatus was filled with water until all the air pockets had been removed.

Fish were acclimatized to the respirometer before experiments for a minimum period of 18 h at a water velocity of 0·3 ft/sec to allow recovery from handling (Black, 1957; Smit, 1965). Mild electric shocks delivered by the rear grill when touched by the fish’s tail were used to train fish to swim steadily at different speeds. Water velocity was increased in steps of 0·2 ft/sec during 2−3 min periods up to a maximum at 2·75 ft/sec and maintained for 5−50 min. When fish would no longer swim the water velocity was reduced to 0·3 ft/sec and in a few cases 24 h allowed for recovery before the experiments were repeated.

To avoid damage or displacement of the electrodes fish were anaesthetized in the respirometer before removal. The exact locations of the electrodes were determined by dissection of frozen animals.

Fish maintained position against low water velocities by frictional contact with the base of the chamber, small amplitude figure-of-eight oscillations of the caudal fin, and slow paddling with the pectorals. At these low speeds slow tail oscillations occurred rarely and only when the fish lost contact with the bottom. Steady swimming occurred at a specific velocity of 0·5−1·5 L/sec that varied considerably from one fish to another and probably reflects a differing degree of contact with the base of the chamber. (Specific velocity was calculated by dividing the measured water velocity by the body length, L, of the fish.) The maximum swimming speed recorded in these experiments was 2·75 L/sec.

As Webb (1971b) reported previously, fish swam restlessly for several minutes after an increase in water velocity, even when executed slowly, before steady swimming was resumed. Fish appeared reluctant to change from one rhythmical pattern of behaviour to a faster one but readily changed to a slower one. At the higher speeds fish would sometimes struggle violently before settling down to swim steadily. At the highest speeds periods of steady swimming were interspersed with bursts of violent struggling before the fish became exhausted and was unable to swim off the electrified grill.

Frequency of tail beats. The results given in Fig. 1 indicate that a linear relation exists between the specific swimming speed and the frequency of oscillation of the tail. Each point plotted is the average frequency of 10−20 consecutive tail beats in a period of steady swimming. The straight line drawn through the amalgamated results in Fig. 1 (f) has a slope of 1:1·57 and does not pass through the origin; it transects the abscissa at a frequency of 1·01 beats/sec. At low speeds the overall propeller efficiency is low, the energy cost high (Webb, 1971 b) and the fish swim using their pectoral fins.

A considerable variation in tail-beat frequency at a given specific swimming speed is indicated in Fig. 1 (c) − (e ). Some of this variation was shown to result from changes in the drag imposed on the fish by the Polythene tubing sutured to its back and leading out of the respirometer. Slight variations in the swimming performance of the fish may also be responsible for some of the scatter (see Bainbridge, 1958).

Amplitude of tail beats. Observations indicated that tail-beat amplitude increased with frequency to a maximum at 3−5/sec. This aspect of the swimming performance of trout has been more extensively studied by Webb (1970, 1971 a) who showed that specific amplitude increased to a maximum of 0·175 as the frequency approached 5/sec as in other fishes of the same body form. Greater lateral excursions of the tail occurred when fish accelerated than were found during steady swimming. During struggling large-amplitude slow movements of the tail were evident.

The muscles of the majority of fish are clearly separated into red and white - the bulk of the musculature being composed of white fibres. In a few species the bulk of the musculature is composed of a mixed population of red and white fibres and is said to have a ‘mosaic’ structure (Boddeke et al. 1959). The muscles of rainbow trout (Salmo gairdneri) are organized in this manner (Webb, 1970).

Since this study is concerned with the role of the muscles in swimming, the activities of the red and mosaic muscles are related in the first instance to their direct mechanical expression - the number of tail beats/second. This is done because of the variation found in tail-beat frequency at a given water velocity (see above). The velocity of the fish is then derived from its tail-beat frequency according to the linear relation (A) drawn in Fig. 1 (f).

Electrical activity was recorded from the red muscles at all swimming speeds at which tail oscillations occurred. Activity on one side of the fish appeared in phase with tail oscillations and 180° out of phase with activity on the opposite side at the same level (Figs. 2, 5 and 6).

Changes in the e.m.g. records obtained at different tail-beat frequencies are seen in Fig. 2. Fig. 3 illustrates the existence of a linear relation between tail-beat frequency and the duration of the interval between bursts. However, burst duration decreased with increasing frequency to a minimum at 3·5−5·0 beats/sec and then increased. Points for measured burst duration against frequency coincide closely with those calculated from the linear relation for interburst interval (Fig. 3). Each point plotted is the average of 13−21 measurements made in a period of even swimming.

At a given swimming velocity each burst of electrical activity is a unique event - the number of potentials varies and the maximum amplitude is determined by different degrees of summation (Fig. 4). The contribution to the waveform by repetitive excitation of one or more motor-units is not certain but it seems probable that a varying number of motor-units are involved. Repetitive firing of motor-units is expected since the red muscles of teleosts have been shown by Barets (1961) and Takeuchi (1959) to be similar to the slow muscles of frog where the tension developed is related to the stimulation frequency.

Potentials in a burst decreased in number with the duration of the burst and increased in amplitude. They generally did not exceed 175 μV (half the diphasic excursion) except rarely when potentials up to 600 μV were recorded. The duration of these potentials was approximately 5 msec. They therefore appear to be faster than junctional potentials (see Roberts, 1969b; Takeuchi, 1959), but slower than typical action potentials from the white muscles (Hudson, 1969).

The amplitude and duration of potentials recorded diphasically vary according to electrode configuration, temporal excitation of different motor-units, and the number, size and spatial distribution of the active motor-units relative to the recording site.

These factors account for the different degrees of summation observed and, further, do not allow the precise origins and nature of the potentials to be assessed.

Bone (1966) considered the potentials he recorded from the red muscles of dogfish to arise from the motor-nerve plexus. In contrast, Roberts (1969 b) concluded, following the application of d-tubo-curarine chloride and suxamethonium to the muscles, that potentials recorded from dogfish dorsal fin red muscles originated in the muscle fibres. Thus it seems probable that the potentials recorded in the present study are myogenic in origin.

Red muscle activity was not inhibited nor augmented when the underlying mosaic muscle mass was active during vigorous swimming (see Figs. 5, 6). Electrodes situated in the red muscle did not record the activity of the mosaic muscle.

At swimming velocities where the tail beat at 3·05−3·60/360 slight and mostly regular activity was recorded from the mosaic muscle in phase with red muscle activity on the same side of the fish (Fig. 5). With increasing frequency this activity increased slightly in amplitude and became completely regular. During periods of steady swimming the activity remained of low amplitude and variable waveform, and resembled that recorded from the red muscles. Faster potentials of considerably greater amplitude than recorded from the red muscles were recorded in association with changes in a movement pattern, for example one-beat accelerations and increases in tail-beat amplitude (Fig. 6). During vigorous swimming and struggling there was an enormous increase in activity of the mosaic muscles (Fig. 6C). In a few instances fish swam at high velocities by a vigorous acceleration followed by a glide and a period of steady swimming during which the fish swam at slightly less than the water velocity (Fig. 6A). This sequence was repeated at intervals of 1−6 sec. Activity was recorded from the mosaic muscles during the steady swimming period. Almost simultaneous discharges were recorded from all the muscles when fish accelerated rapidly (Fig. 6 A).

Responses were not always recorded from the mosaic muscles and when present in many cases diminished and disappeared or left slow potential waves indicative of a loss of integrity of the recording site. Damaged tissue and injury tracts caused by movement of the electrodes in the muscle as a result of flexures of the body during swimming were found on dissection. The damage was related to time and exercise. Red muscle recording sites were not affected in this way.

The present results have revealed the existence of a linear relation between swimming velocity and tail-beat frequency at frequencies up to 5/sec. The amplitude of each tail beat was found to increase with frequency to a maximum at 3−5/sec. The swimming of fish is predicted by the equation given by Bainbridge (1958), where V is velocity (cm/sec), L is the body length (cm) and f is the number of tail beats/sec. The curve relating specific velocity and tail-beat frequency was found by Bainbridge to be linear above 5 beats/sec for three species of fish. Below 5/sec the curve diverged from linearity. Bainbridge further showed that the distance travelled per beat increases (a) with frequency up to a maximum at about 5/sec and is constant thereafter, and (b) linearly with amplitude which also increases with frequency up to about 5/sec and then remains constant. Thus at frequencies less than 5/sec the increase in velocity with frequency would be expected to follow a curve. This contrasts with the present findings. Although Webb (1971a) found that amplitude reached a maximum at a frequency approaching 5/sec he concluded that in the case of trout Bainbridge’s relation could be extended down to a frequency of 2·5 tail beats/sec as the variation in amplitude was slight.

The linear speed/frequency relation revealed in this study has a slope of 1:1·57. The same relation at higher frequencies was shown by Bainbridge, on other fish, to have a slope of 1:1·33. The difference between these slopes may lie within the divergence from linearity at frequencies less than 5/sec described by Bainbridge. As stated above, the linear relation between frequency and velocity described by Bainbridge’s equation can be extended down to 2·5/sec for trout. More probably therefore the difference results from failing to apply a correction factor to the values for velocity. A calculated correction of 1·18 applied to the present measurements of velocity would bring the relation between f and L/sec into agreement with that given by Bainbridge. The measurement of velocity in the Brett respirometer is subject to a significant error due to solid blocking (Webb, 1970, 1971 a). Solid blocking occurs when a body placed in a stream causes an increase in velocity due to a reduction in cross-sectional area. Webb reported an error due to this cause of 7·5−15 % depending on the size of the fish.

No information was obtained in the present series of experiments concerning the maximum sustainable cruising speed or the critical swimming speed (V_c) (Brett, 1964, 1967). The critical swimming speed of trout from the same stock was shown by Webb in one series of experiments to be 1·73L,/sec (Webb, 1971a) and in another 2·oL/sec (Webb, 1971a). The difference was accounted for in terms of seasonal variations. Both of these values are considered to be low (Webb, 1971a) by comparison with the cruising speeds of other salmonids which vary from 1·4 to 4·3 L/sec (see Table 7, Webb, 1971a) and are generally greater than 2·oL/sec. Out of a group of 4 sockeyes, tested at 15 °C for prolonged swimming, one swam at 4·8L/sec for 5520 min, and another swam at 3·7L/sec for 5800 min (Brett, 1964). Amongst the marine teleosts, coalfish, Gadus virens, sustained continuous swimming at 2·1 and 3·oL/sec for 42 days, after which the experiments were stopped, at 10·9 °C (Greer-Walker, 1971, personal communication). The critical velocity is maximal at 15 °C, being less at higher and lower temperatures (Brett, 1964, 1967) and is dependent on length. In the case of sockeye salmon for example V_c decreased from 6·65L/sec for fish 7·74 cm long to 2·65L/sec for fish 53·9 cm long (Brett, 1965). Variations in the physiological condition of different fish influence their critical swimming velocity and account for some variation in performance (Bainbridge, 1962; Brett, 1967). As Webb (1971a) points out, the poor swimming performance of trout obtained from the Nailsworth hatchery reflects their life history, rearing and genetic endowment. Finally, motivation to sustain a particular velocity under experimental conditions is most probably lower than in the wild state (see Bainbridge, 1962).

In this study activities of the muscles have been related to tail-beat frequency and thereby velocity. On the basis of the speed/frequency relation described here (curve (A)Fig. 1 (f)) the same cruising speed is reached at 3·75 beats/sec. The true cruising speed of rainbow trout is reasoned above to be greater than 2·00L/sec. The critical swimming speed is therefore achieved at a tail-beat frequency exceeding 3·75/300 and more probably 4·0−4·5 beats/sec.

Red muscle is active at all swimming speeds. The duration of the electrical discharges decreased with increasing velocity, was minimal at 3·5−5·0 beats/sec and then increased at greater frequencies. This reversal frequency coincided with tail-beat frequency at the maximum cruising speed which suggests the two phenomena are related. At high velocities, during struggling, and when the fish is accelerating considerable activity is recorded from the mosaic muscles. These results are clearly in agreement with those of Bone (1966), Rayner & Keenan (1967) and Roberts (19696) and confirm the functional dualism of the red and white muscles also deduced on other evidence (see Introduction). However, the results presented here indicate that the mosaic muscle in trout becomes active at 3·05−3·60 tail beats/sec. The maximum cruising speed of this fish occurs at a tail-beat frequency in excess of 3·75/sec. The mosaic muscle is therefore also involved at intermediate swimming velocities and in the higher range of cruising speeds. This conclusion supports Webb’s (1970) earlier finding that the mosaic muscle is recruited at 80 % critical swimming velocity. A greater proportion of the mosaic muscle is involved at slightly higher velocities when fish swim by one or two vigorous tail flips followed by a glide and a period of even swimming cyclically repeated. Swimming behaviour of this nature corresponds to steady swimming as defined by Brett (1964) and has been observed amongst coral reef fish in the field (R. C. L. Hudson, unpublished observations) and haddock, Melanogrammus aeglefinus (C. S. Wardle, personal communication).

The mosaic muscle of the salmonids consists of a mixed population of red and white fibres. Any interpretation of the e.m.g.s recorded from this muscle is therefore liable to be ambiguous. Webb (1970) considered the red fibres in mosaic muscle and the lateral line strip to be structurally and functionally similar, although the density of the succinic dehydrogenase sites and fat content were less in the red fibres of the mosaic muscle. He concluded that potentials recorded from the mosaic muscle at 80−100% V_c represented the activity of red muscle fibres. It is proposed here that the white, or mosaic, muscle mass in fish should be treated independently from the lateral fine red muscle, and that this muscle mass, whether containing red fibres or not, is involved at intermediate swimming speeds and during cruising.

Separation of the red and white muscles into two distinct systems appears to occur also at the level of the spinal cord, and it seems probable that activation of these systems involves (i) different levels of excitability of their motoneurone pools, or (ii) different neural pathways. In mammals Henneman and co-workers revealed in an important series of papers (Henneman & Olson, 1965; Henneman, Somjen & Carpenter, 1965; McPhedran, Wuerker & Henneman, 1965; Wuerker, McPhedran & Henneman, 1965) that the participation of a motor-unit in graded motor activity is dictated by the size of its neurone. Henneman & Olson advanced the hypothesis of a size principle that ‘the interrelationship between the functional properties of motorunits and the mitochondrial ATPase content of their fibres depend on the size of the motoneurones that innervate them; the size of the cell dictates its excitability, its excitability determines the degree of use of the motor-unit, and its “usage” in turn specifies or influences the type of muscle fibre required’. Two categories of electrical events recorded from the spinal nerves of curarized dogfish indicate that the white-muscle mass is innervated by larger motoneurones that have a higher threshold to stimulation than those supplying the red muscles (Roberts, 1969a). Rayner & Keenan (1967) found that sodium pentobarbital anaesthesia specifically inhibited whitemuscle activity in the skipjack tuna (teleost).

The white muscles of fish are not considered to consist of a homogeneous population of functionally similar muscle fibres. In mammals the white muscles are now recognized to consist of at least three histochemically distinct types of fibre (Yelfin & Guth, 1970). The white muscles of carp and presumably other fish are similarly subdivided (Ogata, 1958a, b). In trout the red fibres have diameters of 10−80 μm while the white fibres are 50−200 μm in diameter (Webb, 1970). A relationship between histochemical and functional properties and size of a muscle fibre has been demonstrated (Lannergren & Smith, 1966). Although no information is currently available on motor-units in fish it is probable that motor-units of different sizes occur in view of (1) the size principle propounded by Henneman and co-workers, (2) classification of the muscle into three fibre types, and (3) the range of axon diameters in one class indicated by the conduction velocities of 15−24 m/sec (at 20−22 °C) for tench (Barets, 1961) and 17−24 m/sec (at 10−12 °C) for Cottui scorpius (Hudson, 1969) fast motor axons. The whitemuscle mass would therefore appear to be capable of a variety of mechanical actions.

Additional support for the hypothesis that the white (mosaic) muscle mass is involved at cruising speeds is provided by data recently obtained by Greer-Walker (1971). Greer-Walker measured body weight and the diameter of the red and white muscles fibres in unexercised coalfish and in fish exercised continuously for 42 days at 0·93, 2·10 and 3·oL/sec (personal communication). In these fish the red and white muscle masses are distinct. At 0·93L/sec the red fibres alone hypertrophied (55%) consistent with the view that only these fibres are effective at low cruising speeds. At a higher velocity of 2·10L/sec, in contrast, both white and red muscles hypertrophied-increases in diameter of 33 and 61% respectively were found. At speeds of 3·00T/sec increases in diameter of 55 and 70% were found for the white and red muscles respectively. At these higher velocities both red and white muscles are therefore implicated. Further, M. Greer-Walker (personal communication) has found that hypertrophy of the red muscles reaches a peak at about 1·0L/sec and increases slowly thereafter, while hypertrophy of the white muscles starts between 1·o and 2·0L/sec and rises sharply.

It is generally conceded on histochemical grounds that red muscle is well adapted for aerobic respiration with fat as the chief fuel, and that white muscle respires anaerobically utilizing glycogen (Bone, 1966; George, 1962; Webb, 1970). George (1962) points out that in mackerel white muscle is not devoid of fat, mitochondria, lipase, or succinic dehydrogenase, and argues that when subjected to prolonged exercise this muscle could utilize fatty acids derived from extracellular fat stores within the muscle, adipose tissue, and the liver as the energy fuel (see also Brett, 1964). Fatigue of the white muscle during prolonged activity and a dependence on intracellular fat stores would be reduced by the rotation of motor-units. Fish exercised at 0·93L/sec for 42 days showed no weight change, whereas those swimming at 2·10L/sec for the same time and clearly using their white muscles underwent a weight loss of 13 % (Greer-Walker, 1971). Fat depletion of the liver occurred in coalfish following continuous swimming at 3L/sec for 42 days (M. Greer-Walker, personal communication).

This work was supported by a Nuffield Foundation Research Grant awarded to Professor G. M. Hughes which I gratefully acknowledge.