Muscle activation and strain during suction feeding in the largemouth bass Micropterus salmoides

Muscles generate force and strain to power animal behaviors. Muscle function may be considered as the time-course and amplitude of muscle length change (i.e. strain) and force generation in relation to skeletal kinematics and muscle activation (see, for example, Altringham and Ellerby, 1999; Biewener et al., 1998; Finni et al., 2000). Muscle function during a behavior is determined in part by the skeletal movement required by the behavior (kinematic demands), and by the muscle loads incurred during movement (mechanical demands)(Biewener, 2002; Marsh, 1999).

The use of sonomicrometry to measure muscle strain in vivo has vastly increased understanding of muscle function during numerous behaviors,helping to clarify how the specific kinematic and mechanical demands of a behavior affect muscle function (Biewener,2002). Research into muscle function has predominantly focused on locomotor behaviors. Feeding behavior, although of equal evolutionary and ecological significance, has not received commensurate attention. In particular, muscle force and strain during suction feeding have not been studied despite a considerable history of research into the muscular basis of this behavior (Alexander, 1969; Lauder et al., 1986; Osse, 1969).

To successfully capture evasive prey, suction feeding musculoskeletal systems must be able to rapidly expand cranial elements against hydrodynamic pressure gradients of 5–60 kPa(Ferry-Graham et al., 2003; Lauder, 1980b; Muller et al., 1982). Understanding how muscles meet the kinematic and mechanical demands of suction feeding is a necessary first step towards understanding the functional implications of teleost musculoskeletal morphology and may shed light on the biomechanics of suction feeding, which are poorly understood(Ferry-Graham et al.,2003).

To these ends, one aspect of muscle function, the time course and amplitude of muscle fascicle strain in relation to activation and kinematics, was measured in vivo in the sternohyoideus of largemouth bass Micropterus salmoides feeding on evasive prey. The sternohyoideus originates on the pectoral girdle and posterior muscle masses to insert on the hyoid (Figs 1, 2; Edgeworth, 1935; Winterbottom, 1974). Shortening in the sternohyoideus results directly in ventral–caudal rotation of the hyoid and indirectly in lower jaw depression and lateral expansion of the suspensoria (De Visser and Barel, 1996; Fig. 1). Lower jaw depression was also measured to relate strain measurements to relevant skeletal kinematics. This study is the first direct measurement of muscle strain during suction feeding.

The sternohyoideus is known to be active during suction feeding(Grubich and Wainwright, 1997; Lauder et al., 1986; Osse, 1969), but its function cannot be determined without information on fascicle strain patterns. The sternohyoideus may shorten during contraction or may contract isometrically,either to resist action of the epaxial muscles or transmit strain from larger more posterior muscle masses (Fig. 1). Thus, a primary objective of this study was to understand how the sternohyoideus functions during feeding. Because generation of both force and velocity were thought to be important to suction feeding performance, it was predicted that the sternohyoideus would contract at velocities that permit muscular power production during some or all of its active period. In the absence of in vitro data on the velocity at which largemouth bass white-muscle fibers produce maximal power, it was predicted that the muscle would contract at approximately 1/3 maximum velocity or about three muscle lengths per second, assuming a maximum contraction velocity of about ten muscle lengths per second (Askew and Marsh,1998; Hill, 1938; Rome, 1998).

Five largemouth bass Micropterus salmoides Lacapede [28–32 cm standard length (SL)] were used in this study. Largemouth bass are known to specialize on large, elusive prey in the wild and to feed aggressively in captivity despite instrumentation(Etnier and Starnes, 1993; Sanford and Wainwright, 2002). Largemouth bass have also been the subject of numerous studies into the functional morphology of suction feeding(Grubich and Wainwright, 1997; Norton and Brainerd, 1993; Nyberg, 1971; Richard and Wainwright, 1995; Sanford and Wainwright, 2002; Thys, 1997). Fish were fed live goldfish (Carassius auratus) and squid pieces (Loligosp.) prior to experimentation, with feeding discontinued 3–6 days prior to surgery. Animals were maintained at 22°C and were housed and maintained in accordance with University of California at Davis animal use and care protocols (#10211).

Prior to surgery fish were anaesthetized by emersion in 0.3 g l^–1 of MS-222 (tricaine methane sulfonate). Fish were then moved to a surgical tray containing 0.1 g l^–1 of MS-222 and returned to higher concentrations only if they awoke. Surgery lasted less than 30 min and most fish did not awake during surgery.

Sonomicrometry was used to measure gape and muscle strain. In sonomicrometry, small piezo-electric crystals are used to measure distances. The crystals work by transmitting and receiving ultrasound, with the time of sound travel used to measure distance. The utility of using sonomicrometry to measure distances within muscle tissue and water has been established in numerous previous studies (Biewener,2002; Hoffer et al.,1989; Sanford and Wainwright,2002). The speed of sound in muscle was estimated as 1560 m s^–1 (Mol and Breddels,1982). This velocity underestimated kinematic measurements because the velocity of sound in water was overestimated. The timing of lower jaw depression was the parameter of interest not the absolute distance of lower jaw depression, so this overestimation did not seriously affect results.

To measure lower jaw depression, two 2 mm omnidirectional sonometric crystals (Sonometrics Corp., London, ON, Canada) were passed beneath the operculum, rostral to the first gill arch, and into the mouth. One crystal was sutured to the lower jaw just inside the mandibular symphysis, while the other was sutured to the palate lateral to the vomer(Fig. 2). Crystal wires were secured to the dorsal skin of the fish leaving enough slack for normal movement of the jaws and head.

To measure fascicle strain, two 1 mm crystals were inserted between fascicles in the lateral surface of the sternohyoideus. A 1 mm incision was made through the skin and fascia of the sternohyoideus and a 5-0 suture passed through the skin on either side of the incision. Muscle fascicles were gently separated with a blunt probe and the crystal was passed between them with a plastic introducer. The suture was tightened, securing the crystal. The introducer was removed and the skin on either side of the cut was sutured tightly around the wire. Crystals secured by these methods moved freely with underlying muscle fascicles and stayed in place until the end of each experiment. Crystal distance varied from 11–17 mm between preparations. Crystals were inserted along fascicle lines so that fascicle strain would be accurately recorded. The accuracy of crystal distance was checked with calipers prior to recovery of the fish. The movement of crystals with muscle shortening was also checked prior to recovery. Crystal strain and strain rate were interpreted as fascicle strain and strain rate based on the assumptions that crystals were aligned along parallel fascicles and that changes in crystal distance reflected changes in fascicle length.

Bipolar electrodes were fashioned from 0.002 inch (0.051 mm) bi-filament stainless steel, Teflon-coated wire (California Fine-Wire, Grover Beach, CA,USA) by exposing and spreading the wire tips. The electrode wire was bent into a hook 3–5 mm before the exposed tip and the whole wire (approximately 2 m) was threaded through a 26-gauge hypodermic needle. After crystal implantation the electrodes were inserted into the sternohyoideus body via the hypodermic needle and kept in place by the hook in the wire. Care was taken to consistently place electrodes between and just under crystals.

Surgery took place the night before experimentation, allowing the fish to recover overnight before experimentation. Data collection took place the next morning as soon as the fish were willing to feed and continued until satiation. During data collection fish were fed live goldfish (approximately 4 cm SL). At the conclusion of data collection fish were re-anesthetized and instrumentation was removed. Surgical wounds were extensively cleaned with commercial povidone–iodide solution and left open to prevent infection. Fish were returned to their home tank after instrument removal and monitored for 6–8 h. All fish survived experimentation.

Crystal signals were digitized with a Sonometrics TRX-8 conversion box(Sonometrics Corp.) connected to a PC running Sonoview software (Sonometrics Corp.). The PC and software were also used to collect, digitize and record amplified EMG signals. EMG signals were amplified with a differential amplifier (A-M systems, Everett, WA, USA). Using the same software to record EMG and crystal data allowed more convenient synchronization of activity,strain and kinematic signals, but required the EMG signal to be digitized at sampling rates of no more than 500 Hz. One set of data was collected at 700 Hz, but this higher sampling rate markedly increased noise in the strain signal.

Crystal and EMG data were converted to tab-delimited ASCII text files and opened as Microsoft Excel spreadsheets. Crystal data were first smoothed by manually removing obvious outlying points, then with a 3-point rolling average. These methods were found to produce smooth strain profiles without altering the relative timing of discrete events at the sampling frequencies used in this study. The distance between the lower jaw and the neurocranium was normalized by subtracting the minimum distance for each strike from each value and then dividing by the maximum distance for the individual. Fascicle length at each point in time was normalized by subtracting each value from length at onset and dividing by the resting length for the preparation. Because crystal distance at feeding onset varied between strikes, resting length was calculated as the mean crystal distance at onset for each strike in the preparation. The timing of kinematic events (slow onset of lower jaw depression, fast onset of lower jaw depression, peak gape and onset of lower jaw closing), strain events (onset of shortening, onset of fast shortening,peak strain and return to resting length), and activation events (onset and offset) were taken manually from kinematic, strain and activation profiles graphed in Excel. The timing of peak gape was not taken as the time of maximum gape but rather as the end of fast opening(Fig. 3). The onset of fast fascicle shortening was identified by increases to shortening velocities of–1.5 or greater fascicle lengths (FL) s^–1. Although this was an arbitrary value, it was found to be a useful way to demarcate the onset of rapid shortening. The timing of each event was subtracted from each other event to produce a matrix of latencies. Latencies were averaged for each individual and individual means were averaged(Table 1). For events thought to be simultaneous, the latencies were compared to a predicted mean of zero in single t-tests. Events were considered simultaneous if their latency did not significantly differ from zero (P<0.05).

Instantaneous shortening velocity was calculated at each point by subtracting strain at the present time from strain at the previous point and dividing by the time between samples. Two velocities were calculated for each strike: total shortening velocity (the normalized total amplitude of shortening divided by total time to peak strain) and fast shortening velocity(the normalized change in amplitude between the onset of fast shortening and peak strain divided by the latency between these two events).

There was a consistent relationship among measured kinematic, strain and activation patterns (Figs 3, 4; Table 1). Lower jaw depression began with a slow phase of small amplitude, which was followed by a fast phase of large amplitude ending in peak gape(Fig. 3). The fast phase averaged 26±1 ms in duration. The lower jaw remained depressed for an average of 28±5 ms before the onset of closing and closed in an average of 44±3 ms (Table 1). The kinematics of lower jaw depression were as described in other studies of similarly sized M. salmoides(Richard and Wainwright, 1995; Sanford and Wainwright, 2002),indicating that instrumentation did not gravely affect performance in this study.

Sternohyoideus activity was observed in all feeding events, but the timing of activation varied among feeding events. For instance, in some strikes activity was nearly simultaneous with the onset of shortening(Fig. 3B), while in others sternohyoideus shortening began 40–60 ms before onset of activation(Fig. 3E). On average,observable activity began 23±6 ms after the onset of sternohyoideus strain. The sternohyoideus was always active before the fast shortening period. The latency between onset activation and onset of fast shortening was less variable within individuals than the latency between onset of activation and onset of slow shortening (t-test of latency s.e.m. values by individual P<0.01, N=5). Activity in the sternohyoideus usually ended simultaneously with peak strain and the onset of lower jaw closing(Fig. 4, Table 1). Sternohyoideus activity was never observed prior to the onset of muscle shortening, but was always observed before the onset of fast shortening(Fig. 4, Table 1).

Sternohyoideus fascicles shortened by an average of 11±0.7% across individuals. The onset of shortening was nearly simultaneous with the onset of slow jaw depression (Table 1),and typically continued smoothly until maximum strain was reached(Fig. 3). In two fish it was common for the sternohyoideus to discontinue shortening or lengthen after the onset of fast lower jaw depression (Fig. 3A,C,D). The sternohyoideus usually began to shorten again about the time of peak gape. The relationship between sternohyoideus shortening and peak gape varied among individuals. In three individuals there was significantly greater strain after peak gape than before peak gape(Fig. 5); in the other two individuals there was roughly equivalent strain before and after peak gape. In all strikes the sternohyoideus continued to shorten after peak gape. On average, peak strain occurred 28±3 ms after peak gape. The onset of lower jaw closing was simultaneous with the cessation of fascicle shortening and muscle activity (Figs 3, 4; Table 1). The return of sternohyoideus to its resting length was simultaneous with the return of the jaw to its closed position (Table 1).

Averaged velocity profiles for three individuals are shown in Fig. 6. Shortening started slowly but increased rapidly near the time of peak gape. This rapid increase in negative slope of the strain profile marked the onset of the fast shortening period. Shortening velocity during fast lower jaw depression was variable between individuals. In most individuals the sternohyoideus contracted rapidly during fast lower jaw depression, but in two individuals the sternohyoideus showed little strain or positive strain during most of rapid lower jaw depression (Figs 3, 6). In all individuals sternohyoideus shortening was faster after peak gape than before peak gape(Fig. 5). Instantaneous shortening velocities during the fast portion of shortening ranged from–2 to –5 FL s^–1(Fig. 6). Fast shortening levels were typically maintained until peak strain. Total velocity of fascicle shortening for the sternohyoideus averaged –1.3±0.12 FLs^–1 across all individuals. Total fast shortening velocities averaged –2.5±0.40 FL s^–1(Fig. 7). In all strikes the majority of peak strain occurred during the fast period(Fig. 7).

The sternohyoideus has been implicated as a suction feeding actuator because of its anatomical position, and because it typically depolarizes during suction feeding behavior (Grubich and Wainwright, 1997; Lauder et al., 1986; Osse,1969). While the presence of activity indicates that the sternohyoideus generates force during suction feeding, no previous study has shown that the sternohyoideus actually shortens during suction feeding. Muscles may lengthen, shorten or remain of constant length when active,depending on the stress they produce and the stress under which they are placed. Only by shortening can a muscle power skeletal kinematics, but shortening, especially rapid shortening, decreases the amount of force produced by the muscle according to the force–velocity relationship observed in skeletal muscle (Hill,1938). If the sternohyoideus maintained constant length or lengthened while active, its role might be to transmit strain from the rotation of the pectoral girdle relative to the hyoid, allowing larger, more posterior muscle masses to generate strain. If this were true its function would be analogous to the transmission of strain by posterior muscle masses in swimming fish (Altringham and Ellerby,1999; Wardle et al.,1995). Such rotation occurs when the neurocranium is rotated dorsally by the epaxial muscles, pulling the hyoid away from the pectoral girdle, and when the pectoral girdle is rotated caudally by the hypaxial muscles (Aerts et al., 1987; Osse, 1969; Thys, 1997). The active shortening observed in the sternohyoideus indictates that it, in fact,actuates suction feeding kinematics.

The sternohyoideus consistently shortened after peak gape(Fig. 5). In three of the five individuals more strain was observed after peak gape than before. This pattern is consistent with the multiple functions of the sternohyoideus, including lower jaw depression, hyoid depression and lateral expansion of the suspensoria (Fig. 1). Suction feeding fish in general, and largemouth bass in particular, exhibit an anterior-to-posterior wave of expansion during suction feeding. Peak gape precedes hyoid depression, which precedes peak opercular expansion(Lauder, 1980a; Osse, 1969; Sanford and Wainwright, 2002). This kinematic pattern is possible because the linkage between the proximal hyoid and lower jaw allows the hyoid to continue depression after the lower jaw has reached its ventral limit (Fig. 1). Shortening after peak gape would continue ventral–caudal rotation of the hyoid, resulting in continued depression of the buccal floor and continued lateral expansion of the suspensioria.

In all individuals the sternohyoideus increased shortening velocity after peak gape (Fig. 5). This pattern suggests that sternohyoideus loading may be reduced after peak gape. Loads may be reduced because the lower jaw is no longer rotating, and because other cranial structures are not accelerating as rapidly as prior to peak gape, reducing hydrodynamic load (Sanford and Wainwright, 2002). The major source of loading during suction feeding appears to be negative internal buccal pressure(Alexander, 1969), the magnitude of which is greatly decreased after peak gape(Sanford and Wainwright,2002). These factors would all contribute to decreased loading and increased shortening velocity after peak gape.

In two of the five studied individuals the sternohyoideus lengthened slightly or contracted isometrically during fast lower jaw depression. One explanation for this observation is that the sternohyoideus was stretched by rotation of the hyoid away from the pectoral girdle as described above. By contracting eccentrically the sternohyoideus might be able to generate greater force than it would be able to during shortening(Hill, 1938; Lindstedt et al., 2002). Why this pattern was consistently observed in some individuals and not others is not apparent from this study, but it may have been the result of subtle differences in muscle activation between individuals. Individual variation is common in studies of both suction feeding and muscle function (e.g. Gillis and Biewener, 2000; Grubich and Wainwright, 1997). Furthermore, jaw opening in teleosts is actuated by multiple complex and redundant musculoskeletal systems (Adriaens et al., 2001; Lauder,1980a; Westneat,1990), so there is potential for variation in how these systems are used between individuals.

In many reported vertebrate behaviors muscle activation precedes shortening by 20–80 ms (e.g. Franklin and Johnston 1997; Gillis and Biewener, 2000; Olson and Marsh, 1998; Lutz and Rome,1994). In this study measurable activation was not detected until an average of 23 ms after the onset of the slow fascicle shortening(Table 1, Fig. 4). The sternohyoideus may be returning to resting length after being passively stretched by an antagonistic muscle or be slackening by action of a synergistic muscle. Elevation of the lower jaw by the adductor mandibulae would stretch the sternohyoideus, but these muscles are not detectably active prior to expansion in largemouth bass (Wainwright and Richard, 1995).

Passive shortening in the sternohyoideus might also have been caused by depression of the lower jaw through the opercular linkage(Lauder, 1980a). A small muscle, the levator operculi, links the operculum to the neurocranium and is thought to depress the lower jaw through a linkage involving the opercular series. This muscle is typically active prior to the sternohyoideus(Grubich and Wainwright, 1997; Lauder et al., 1986; Osse, 1969) and is thought to actuate the slow phase of lower jaw depression(Aerts et al., 1987; Osse, 1969). In doing so it may allow the sternohyoideus to shorten. The onset of shortening and the onset of activation were variable within individuals, but the sternohyoideus was never active prior to shortening. However, several strikes showed nearly simultaneous onset of shortening, jaw depression and activation(Fig. 3). In these strikes the levator operculi may have been activated at the same time as the sternohyoideus.

The shortening velocities measured in this study suggest that the sternohyoideus functions to generate mechanical power during suction feeding. In all individuals, the sternohyiodeus actively shortened during part or all of its active period, generating mechanical power. Shortening velocity averaged –2.5 FL s^–1 during fast shortening. Although it cannot be concluded from in vivo data alone, this velocity may maximize mass-specific production of muscle velocity. Peak muscular power production occurs at approximately 1/3 maximal velocity(Askew and Marsh, 1998; Hill, 1938). Maximal velocities for suction feeding muscles are not known in fish, but maximum velocities for fish white axial muscle at 22° approximate 7–10 FL s^–1 (Aerts et al., 1987; Beddow and Johnston,1995). Furthermore, the shortening velocities measured in this study were similar to those of other ectothermic vertebrates performing power-limited behaviors at similar temperatures(Ellerby and Altringham, 2001; Lutz and Rome, 1994; Nelson and Jayne, 2001; Olson and Marsh, 1998). Without in vitro data, however, it cannot be concluded that the sternohyoidues contracts at velocities that permit maximal mass-specific power production.

The power required for suction feeding would therefore increase linearly with decreasing duration of cranial expansion(t_final–t₀) even if the incurred hydrodynamic resistance were independent of strike duration. In fact,hydrodynamic loading probably increases with increasing speed of volume expansion. Although the hydrodynamics of suction feeding are extremely complex and poorly understood, the pressure measured inside the expanding buccal cavity probably results from the inertia of water entering the mouth(Muller et al., 1982); it can therefore be expected that the more rapidly the volume expansion takes place the more intense the negative pressures that resist expansion. There is empirical evidence to support this assumption: Sanford and Wainwright(2002) found that internal pressure in largemouth bass increased with increasing acceleration of cranial expansion.

Similarly, positive pressures outside expanding cranial elements may increase with increasing suction feeding speed. This pressure is due to the inertia of water (added or virtual mass) displaced by the expanding cranial elements and would be expected to increase linearly with increasing cranial acceleration (Batchelor, 1967; Daniel, 1984; Vogel, 1994). In a detailed theoretical analysis of shrimp tail flips, which occur on time scales similar to suction feeding, Daniel and Meyhofer(1989) found that predicted muscle force increased exponentially with decreasing duration of the tail flip. Furthermore, differential pressures measured on the tail of swimming cod Gadus morhua were shown to increase with increasing tail rotational acceleration (Webber et al.,2001).

If the above analysis is correct, increased suction feeding performance, in an individual fish, is limited by muscular power. Therefore, fish that use suction feeding to capture evasive prey ought to possess musculoskeletal morphologies that permit fascicles to shorten at velocities that maximize their intrinsic power capabilities. In other words, fish that feed on evasive prey ought to be `geared' for power generation. This study supports this theoretical prediction: sternohyoideus fascicles contract at velocities consistent with the production of mechanical power. Whether the shortening velocities measured in this study approximate the optimal shortening velocity for power production must be determined by in vitro measurements of muscle fiber performance, as has been done in other systems(Lutz and Rome, 1994; Marsh et al., 1992; Peplowski and Marsh, 1997; Rome et al., 1993; Wakeling and Johnston,1998).

I would like to thank Miriam Ashley-Ross and Richard Marsh for help with techniques, Rachel Wiley for help during experiments, and Peter Wainwright,Robert Josephson and two anonymous reviewers for their invaluable help with the manuscript. Some equipment used in this study was purchased with funds from the Jastro-Shields Graduate Research Grant, and a University of California, Davis Graduate Research Award. This research was also supported by NSF grant IBN-0076436 to Peter Wainwright.