Spatial Variation in Epaxial Muscle Activity During Prey Strike in Largemouth Bass (Micropterus Salmoides)

What is the functional unit of the fish axial muscle system? Owing to the complicated shape, fiber orientation, myoseptal connections and innervation patterns of the fish myomere, this question has remained unanswered. Recent work by Jayne and Lauder (1995) has demonstrated through electromyography that, while all portions of the fish myomere may be activated synchronously during an escape response, some portions remain inactive during other forms of locomotion. The present study investigates whether the dorsalmost portion of the fish axial muscle system can operate independently of the more medial portions during the specific behavior of prey strike.

Several electromyographic studies have established that cranially attached, epaxial muscle is active during the dorsocaudal, neurocranial rotation that occurs in fish prey strike (Osse, 1969; Liem, 1973; Liem and Osse, 1975; Lauder, 1980; Wainwright and Lauder, 1986; Westneat and Wainwright, 1989). However, no study to date has determined whether the entire epaxial muscle mass is activated during prey strike or whether only discrete myomeric regions are recruited. Since neurocranial rotation is one of the components of the inferred ancestral prey strike of ray-finned fishes (Lauder, 1980, 1982), understanding the role epaxial muscle plays in head rotation may provide insight into the relationship between evolutionary changes in vertebrate feeding kinematics and epaxial muscle structure.

In the present study, findings from musculo-skeletal dissections are used to develop hypotheses of white muscle activity during prey strike in two discrete epaxial myomeric regions: the extreme dorsal region (the epaxial arm region) and the more medial region located closer to the main horizontal septum (the epaxial cone region). These regions have been described in scombrids by Westneat et al. (1993) and are represented schematically in Fig. 1. The arm and cone epaxial muscle regions are morphologically distinct; the arm region consists of adjacent, slab-like sections of muscle while the cone region consists of conical, internested sections of muscle. The two regions also have different overall muscle fiber angles; the muscle fibers of the arm region are oriented nearly parallel to the longitudinal body axis while the muscle fibers of the cone region follow more helical trajectories (Alexander, 1969). The longitudinally oriented muscle fibers of the arm region appear better situated mechanically to produce a dorsal rotation of the neurocranium upon shortening than do the more helically oriented fibers of the medially located cone region. Furthermore, the arm region is the major part of the axial myomeric muscle that attaches to the cranial occiput, whereas the more medially located cone region does not exhibit as direct a connection to the skull. The present study tests the possibility that these distinct differences in overall fiber angles, insertion areas and morphology, as revealed by dissection, correlate with differences in the activity patterns of these two regions during prey strike.

Three hypotheses, formulated through extensive epaxial muscle dissections of five individuals, are proposed and tested through electromyography. (1) The dorsalmost region of the epaxial muscle (the arm region) which has nearly longitudinally oriented muscle fibers will play a more prominent role in prey strike than the medial cone region which has more helically oriented muscle fibers. (2) The anteriormost portion of the arm region which has visible tendinous connections to the neurocranium will show greater activity levels than the more posterior portion of the arm region during prey strike. (3) Activity of the arm region will not be obligatorily correlated with activity of the medially located cone region during prey strike. This third hypothesis tests the assumption that myomeres (discrete, serially repeated muscle segments that comprise the axial muscle) operate as obligatory functional units during the behavior of prey strike.

The largemouth bass Micropterus salmoides Lacepède, a freshwater fish indigenous to eastern and central North America (Etnier and Starnes, 1993), was used in the present study.

Morphological observations were made on specimens from the Carolina Fish Hatcheries, Faison, NC, USA, and from ponds in Carrborro and Durham, NC, USA. Electromyography was performed on specimens collected from Lake Jackson, Leon County, Tallahassee, FL, USA. Although populations of largemouth bass from peninsular Florida are recognized as a valid sub-species, M. salmoides floridanus, by Bailey and Hubbs (1949), comparative dissections revealed no quantifiable differences in the overall axial muscle structure of the North Carolina and Florida populations sampled in the present study.

Epaxial muscle dissections were performed on five fresh specimens of adult M. salmoides for the purpose of describing the morphology of the epaxial muscle and its connections to the cranium. Fish ranged in size from 210 mm to 350 mm in standard length (SL). Dissections were performed on both the right and left sides of the specimens. Two types of dissections were performed. (1) Skinning the lateral left side was followed by lateral dissection to reveal the shape of the epaxial myomeres and to permit study of the tendinous and muscular attachments to the cranium, axial skeleton and skin. Particular attention was paid to epaxial attachments to the supraoccipital crest of the skull and epineural bones. (2) A dorsal incision along the body midline was followed by gentle separation of the dorsalmost epaxial muscle away from the axial midline in order to reveal tendons on the medial face of the anteriormost axial muscle.

To investigate the possibility that the epineural bones are moved posteriorly during prey strike, X-ray photographs (Hewlett Packard Faxitron X-ray machine, KVP 30, exposure time 20 min) were taken of a freshly killed largemouth bass (SL 281 mm). The first X-ray photograph was taken of the bass with its mouth closed. The second was taken with the neurocranium rotated and pinned at an angle of approximately 8 °. Largemouth bass of similar size have been reported to achieve 30 ° of neurocranial rotation during prey strike (Richard and Wainwright, 1995). Thus, an angle of 8 ° represents a conservative approximation of a rotated head position during prey strike. The cranial outline and anterior axial skeleton were traced from both X-ray photographs and the rotation angles measured. In addition, the distances between the supraoccipital crest, epineural bones and first dorsal fin support (dorsal pterygiphore) were measured.

Four specimens of largemouth bass were obtained from Lake Jackson in Leon County, Tallahassee, FL, USA. Individuals 1–4 measured 261 mm, 330 mm, 170 mm and 158 mm SL, respectively. All specimens were fed a maintenance diet of Carassius sp. (goldfish) and Poecilia latipinna (sailfin mollies) and were kept in 30–100 l tanks at room temperature (21–23 °C).

To document patterns of muscle activity during prey strike, electromyograms were recorded using fine-wire, bipolar, stainless-steel electrodes constructed from equal lengths of paired, 0.051 mm diameter, insulated wires that were glued together such that the distance between the recording tip was fixed. Electrode tips, approximately 0.6 mm long, were exposed by removing the insulation using a razor blade under a microscope and were mounted in hypodermic needles with the tips bent back to form anchoring hooks (Basmajian and Stecko, 1962).

Prior to electrode insertion, fish were anesthetized in a solution of buffered tricaine methanesulfonate (MS222) (approximately 0.7 g l⁻¹) until no respiratory movements were visible. During electrode implantation, anesthesia was maintained using a diluted solution of anesthetic (approximately 0.3 g l⁻¹) as water was manually fanned over the gills.

A series of eight electrodes was then implanted subdermally along the left half of epaxial musculature, approximately 8–10 mm deep following the pattern shown in Fig. 1A. Five electrodes (A1–A5) were implanted dorsal to the lateral line in the epaxial arm region of the myomeres. Three of these electrodes (A1–A3) were located in the region of the neurocranial-epineural-epaxial connector tendons (NEC tendons). NEC tendons, defined and described in more detail below, are a set of robust tendons originating on the medial face of the dorsalmost epaxial muscle and inserting on the posterior edges of the neurocranium. The other two arm region electrodes (A4 and A5) were located caudal to the NEC region. Another set of three electrodes (C1–C3) was implanted ventral to the lateral line in the epaxial cone region. The positions of the cone electrodes were matched longitudinally with the arm region electrodes (A3–A5) (Fig. 1). Electrode A1 was placed in the epaxial muscle lateral to the posterior margin of the supraoccipital crest, while electrode A2 was placed posterior to the first electrode between the posterior margin of the skull and the anterior boundary of the first dorsal fin. The next two electrodes (A3, C1) were placed at the level of the first dorsal fin spine, with one electrode occupying a position midway between the dorsal fin anterior base and the lateral line, and the other midway between the lateral line and the main horizontal septum. The next pair of electrodes (A4, C2) was placed at the level of the first dorsal fin’s posterior margin, with one electrode dorsal to the lateral line and the other ventral to the lateral line. The last electrode pair (A3, C3) was located slightly posterior to the anterior base of the second dorsal fin, with one electrode being inserted dorsal to the lateral line and the other ventral to the lateral line. The most caudal region of the fish, posterior to the caudal border of the second dorsal fin, was not tested in this study.

The wires of each electrode were gathered together and glued into a common cable, which was then sutured to the skin on the mid dorsum to limit disturbance of the electrodes due to tension on the cable. Once suturing was complete, fish were resuscitated by directing a stream of fresh water over their gills until they initiated their own gill movements. The entire procedure took less than 20 min per fish. Fish were allowed to recover for at least 2–3 h before testing.

Following recovery, bass were presented with live mollies. Each strike at the live prey was recorded on a voice track simultaneously with the EMG recordings. Between 11 and 17 prey strikes were analyzed for each individual on a 14-channel TEAC XR-5000 FM analog recorder using a tape speed of 9.5 cm s⁻¹. EMG signals were amplified 10 000 times using Grass P-511 preamplifiers, and a bandpass of 100–3000 Hz was employed in addition to a 60 Hz notch filter. Hard copies for visual inspection of the recorded events were produced using a Graphtek thermal-array chart recorder. The voice track recordings as well as actual EMG signals were used to differentiate between swimming events and prey strikes. Burst swimming was induced and recorded in each fish by means of a manual prod in order to confirm the proper functioning of all implanted electrodes.

Recorded electromyographic data were digitized using a custom-designed software system and a sampling rate of 8 kHz. Each EMG burst was displayed on a 0.4 m VGA computer monitor such that 125 ms of data occupied the entire width of the monitor. An on-screen movable cursor was used to mark the times of onset and offset visually, and a custom-designed computer program was used to determine the onset time, the duration and the rectified area (integrated EMG signal area over time) of a single burst of muscle activity. Rectified area is correlated with the amount of tension developed in the muscle sampled (Hylander and Johnson, 1989) and may also reflect an increase in the repetitive rate of activation of individual motor units (Basmajian and De Luca, 1985). The mean amplitude (termed intensity in the present study) of each muscle burst was also calculated by dividing the rectified area by the duration of activity.

After a fish had performed at least 11 strikes, the experiment was terminated and the fish was killed with an overdose of MS222. Electrode locations were determined through post-mortem dissections under a Vickers optical surgery dissecting microscope. Owing to the internested nature of the myomeres (see Jayne and Lauder, 1995), the exact depth of the electrodes greatly affects myomeric position. To ensure proper identification of electrode position, a cross section was made immediately anterior to each electrode. Viewing the electrodes in cross section helped to determine accurately into which myomeric arm and cone each electrode was inserted.

The myomeric location of the electrodes was based on counting the myomeres from anterior to posterior along the superficial portion of the horizontal septum. Owing to the technical difficulty involved in placing electrodes precisely into the arm and cone of the same myomeres, arm and cone matches were made following post-mortem dissections. Three of the epaxial arm region electrodes (A2 or A3, A4, A5) were paired with the three epaxial cone region electrodes (C1–C3) as determined by myomeric number. The separation between the arm region and the cone region electrode pairs was never more than three myomeres or approximately 5 mm. The data set consisted of a total of 12 arm/cone pairs or three pairs per fish.

For each fish, the mean values of onset time, rectified area, duration and intensity of each electrode’s activity bursts for each prey strike were calculated using SAS (1989). Onset times were calculated for posterior electrodes (A2–A5) relative to the most anterior arm electrode site (A1). Using Systat version 5, single-factor analyses of variance (ANOVAs) were performed for each fish on the effect of electrode position (A1–A5) on the onset time and duration of the recorded EMG signals. ANOVAs were also performed to compare (1) activity levels of the epaxial arm region electrodes (A1–A5) with activity levels of the cone region electrodes (C1–C3) and (2) activity levels of the anterior arm electrodes (A1–A3) with activity levels of posterior arm electrodes (A4 and A5). Lastly, Pearson correlations and scatterplots were used to investigate correlations between arm electrode activity levels and adjacent cone electrode activity levels.

Removal of the skin revealed the superficial myoseptal pattern of the anterior epaxial muscle (Fig. 1B). The cranially sloping orientation of the anterior myosepta and the longitudinal fiber angles of the muscle suggest that the myosepta are well positioned for transmitting contractile myomeric forces to the cranium for dorsal rotation of the neurocranium. The slab-like arms of the first three anterior myosepta arch onto the head, and the muscle fibers between these septa are oriented nearly longitudinal to the body axis. The muscle fibers from the most dorsal regions of these first three myomeres fuse along the dorsal midline and, although faint separation of the muscle segments is visible, the segments cannot be separated along these barely visible septal lines without substantial tearing of the muscle fibers. The arm regions of the anterior myomeres connect to the skull via the supraoccipital crest and occipital, epiotic and opisthotic bones in the posterior region of the cranium.

A set of short, robust tendons arises medially from the most anterior epaxial myomeric arm region. These tendons run cranio-ventrally along the sagittal plane of the vertical septum and span the distance between approximately the fifth dorsal pterygiophore and the supraoccipital crest. Distinguishable from the anterior myosepta of the epaxial muscle mass, these tendons arise medially from the extreme dorsal myomeric muscle and insert onto the supraoccipital crest, the posterior edge of the epineural bones, the first three neural spines, the connective tissue between the neural spines and the first five dorsal pterygiophores. These tendons do not ensheath segments of muscle, like typical myosepta, but rather provide a cranial and axial skeletal anchor for the anterior region of the dorsal epaxial muscle. These tendons are defined here as neurocranial-epineural-epaxial connectors (NEC tendons). The posterio-dorsal orientation of the NEC tendons suggests that they may play a prominent role in transmitting epaxial muscle force posteriorly via their insertions onto the supraoccipital crest and the epineural bones. Contraction of the epaxial muscle arm region presumably initiates neurocranial dorsad rotation about the vertebral column. This contraction may also pull the epineural bones posteriorly, thus potentially providing additional rotational space for the neurocranium (Fig. 2).

The potential for posterior epineural movement is verified by comparing tracings from X-ray photographs of a largemouth bass (SL 281 mm) with its mouth closed (Fig. 3A) and with the head rotated and mouth pinned open as if in the middle of prey strike (Fig. 3B). The distance between the supraoccipital crest and the first epineural bone decreases from 2.5 mm in the bass with its mouth closed to 1.6 mm when the mouth is open. The distance between the third epineural and the first dorsal fin support decreases from 6.9 mm (closed mouth) to 5.1 mm (open mouth). Although a manual manipulation of a freshly dead bass cannot reproduce fully the prey strike actions of a live bass, the measurements from the X-ray photographs show that the skeletal elements can move in relation to each other during head rotation.

The results from these dissections suggested that the epaxial muscles that were connected by the robust NEC tendons to the cranium and to the immediately adjoining vertical septal elements (Fig. 1A: electrodes A1, A2 and A3) would demonstrate a larger amount of activity during prey strike, as measured by muscle integrated area and intensity, than those in the posterior arm region located outside the NEC tendon area (Fig. 1A: electrodes A4 and A5). The region of epaxial muscle that is attached directly to the occiput of the skull through robust tendinous connections is in the appropriate orientation to produce dorsal rotation of the skull, while the immediately posterior muscle attached to the epineurals is oriented such that, upon contraction, the epineurals will be pulled posteriorly, thus creating a space in which to accommodate the rotation of the cranium.

All arm electrodes (A1–A5) were significantly more active than the cone electrodes (C1–C3) (Table 1; Figs 4–6), with the most prominent regions of activity occurring in the first three anterior arm electrodes (A1–A3) (Table 1; Fig. 7). For fishes 1 and 4, the anterior three arm electrodes (A1–A3), located in the region of the NEC tendons, had significantly greater EMG intensities than did the posterior two arm electrodes (A4 and A5) (Table 1; Fig. 7).

Activity of the cone electrodes (C1–C3) was variable (Figs 4, 5). The cone electrodes were inactive during 48 % of the prey strikes (mean value for all cone electrodes for all fish) while the anteriormost dorsal electrodes (A1 and A2) were active for 100 % of the total prey strikes (Fig. 5). Electrodes A3 and A4 were active in over 70 % of the total prey strikes for all fishes, and for fishes 3 and 4 the most caudally located dorsal electrode (A5) was active in over 90 % of the total strikes. Fishes 1 and 2 showed activity in electrode A5 in over 40 % of the total prey strikes (Fig. 5).

Onset times were highly variable among the four bass. Fishes 1, 2 and 4 showed a significant correlation between onset time and electrode number (A1–A5), while fish 3 showed no correlation (Table 2). Tukey–Kramer post-hoc tests revealed that for fish 1 the onset time of electrode A5 was significantly different from that of electrodes A1–A4, while in fish 2 the onset times of electrodes A4 and A5 were significantly different from those of A1–A3. For fishes 3 and 4, none of the onset times of the electrodes (A1–A5) was significantly different, even though fish 4 demonstrated a significant overall effect of electrode number on onset time.

Duration time varied significantly with electrode number for fishes 1, 2 and 4 but not for fish 3 (Table 2). In fishes 3 and 4, considerable overlap was found in the activity period of the posterior electrodes (A4–A5) with that of the anterior electrodes (A1–A3) (Fig. 8).

Pearson correlation matrices were constructed to explore correlations in EMG intensity levels between closely matched arm electrodes and cone electrodes in the anterior, mid and posterior regions of the body. In eight out of twelve pairs, no significant correlations were found. For the four cases in which there were significant correlations, scatterplots of arm EMG intensity levels versus cone EMG intensity levels were made. For fish 1, values for pair 3 (A5–C3) were significantly correlated (r²=0.53). In fish 3, values for pair 1 (A3–C1) were significantly correlated (r²=0.65), while for fish 4 pairs 1 (A3–C1) and 3 (A5–C3) both had an r² value of 0.53. The correlations between these variables were not strong. Even in the best of these four cases, pair 1 of fish 3, only 44 % of the variance in the cone intensity values can be explained by the arm intensity values. Activity in a cone electrode, however, was always accompanied by some level of activity in the matched arm electrode, but the reverse was not true – arm electrodes were active in many instances with no corresponding activity in the matched cone electrode (e.g. Fig. 4): pair 1 (A3–C1); pair 2 (A4–C2) and pair 3 (A5–C3) (see also Fig. 5).

Understanding the role epaxial muscle plays in neurocranial rotation – a key component of the fish prey strike – is critical to the understanding of myomeric structure and overall epaxial muscle morphology. While previous electromyographic studies have recorded from only one anterior epaxial muscle site during prey strike (Osse, 1969; Liem, 1973; Liem and Osse, 1975; Lauder, 1980; Wainwright and Lauder, 1986; Westneat and Wainwright, 1989), the present study investigates epaxial muscle activity along the length of the body at several sites. During prey strike, only specific regions of the epaxial muscle were found predominantly to be active (Fig. 4). The dorsalmost region, composed of slab-like, longitudinally oriented muscle fibers, was capable of being active independently of the medial region of conical muscle. Furthermore, activity of the dorsalmost epaxial muscle was not confined to the region directly connected to the neurocranium but extended further posteriorly, at least to the level of the second dorsal fin. During prey strike, epaxial muscle activity appears to be dorsally regionalized, thus disqualifying the fish myomere as the obligatory functional unit of the fish axial muscle system.

The capacity for functional segregation of activity within one myomere was first documented in largemouth bass by Jayne and Lauder (1995). These authors found that, while the arm and cone regions of a myomere are coupled during an escape response, the two regions appear capable of a wide range of activity levels during burst-and-glide locomotion. They also found that the cone region could be active without the arm region being active. One fish in the Jayne and Lauder (1995) study showed unexpected activity in the arm region independent of the cone region, similar to the pattern described above for prey strike; however no functional explanation for this activity was provided. The present study extends the findings of Jayne and Lauder (1995) and demonstrates clearly that specific myomeric regions can be functionally uncoupled. More specifically, the arm region of the epaxial muscle can be active while the adjoining cone region is relatively quiescent during the behavior of prey strike (Fig. 4). Together, these findings open to question the functional significance of the whole-body, helical muscle fiber trajectories described for fish axial muscle by Alexander (1969).

Spatial variation in the function of the fish myomere is associated with morphological diversity within this type of muscle segment. A typical myomere contains regions of slab-like muscle connected to regions of conically shaped muscle, and each of these regions has different overall muscle fiber angles. The muscle fibers of the dorsalmost slab-like region (the arm region) are oriented nearly parallel to the longitudinal body axis, while the muscle fibers of the medial region (the cone region) follow a more helical trajectory (Alexander, 1969). Additionally, one single myomere may span 7–10 intervertebral joints (Jayne and Lauder, 1994).

In three out of four fishes, the most anterior arm electrodes (A1–A3) appeared to play a more prominent role in prey strike than did the more posterior epaxial areas sampled (A4 and A5). The anterior electrodes are in the region of epaxial muscle that connects to the neurocranium through a set of robust parasagittal tendons described and defined in the present study as neurocranial-epineural-epaxial connector tendons or NEC tendons (See Fig. 2). Multiple aponeuroses, tendons and subdivisions within musculature are typical of areas of force concentration where movement is enacted over a wide angle (Dullemeijer, 1974). Similarly elaborate epaxial muscle connections to the neurocranium have been described for a characoid (Lesiuk and Lindsey, 1978), a cyprinid (Howes, 1979) and a luciocephalid (Lauder and Liem, 1981), all fishes believed to be capable of extreme neurocranial elevation during feeding behavior. Of particular interest in the largemouth bass, however, is the finding that epaxial muscle activity during prey capture is not confined only to the area of the NEC tendons. Considerable overlap of activity within the arm region is present (Fig. 8), with the arm electrodes being activated posteriorly to at least the region of the second dorsal fin (Figs 4, 8) and potentially further caudally. This has not been reported previously for any species.

In all fishes, the electrodes in the dorsalmost epaxial region (A1–A5) showed significantly more activity than did the medial cone electrodes (C1–C3) (Fig. 6; Table 1).

Interaction between the activity levels of the arm and cone regions may vary with the strength of the prey strike. For weak strikes, the arm region may be the most commonly recruited epaxial region. For more forceful strikes, the cone region may also be recruited, possibly aiding in the establishment of an anchoring platform for the contracting myomeric arm region. In order to assess more fully the variability seen in the activity levels of the cone region relative to the arm region during prey strike, a method of quantifying the strength of the strike would be useful. One such experiment would be to synchronize high-speed video with EMG recordings and to measure both the degree and speed of neurocranial rotation during the strike. These two measurements could provide a quantitative strength index for each strike which could be used to investigate EMG variability between strikes. Another possibility is that cone region activity may be due simply to bending of the fish to the left as a result of aligning itself to the prey item or to lunging forward during a particular strike. It would be interesting to record from the same cone sites on contralateral sides of the fish during prey strike to distinguish between these two possibilities.

Similar functional segregation within single muscles has been reported for several other vertebrates, including monitor lizard, Varanus exanthematicus, jaw adductor muscles (Smith, 1982) and cat, Felis domesticus, limb muscle (Hoffer et al. 1987). In V. exanthematicus, the differential activity detected within the pterygoideus muscle during inertial feeding was not demonstrated to correspond with gross morphological subdivisions of the muscle (Smith, 1982). The distinct anatomical regions of the cat sartorius muscle (Hoffer et al. 1987), however, do appear to have discrete functions. The anterior region acts to extend the knee and flex the hip, while the medial region flexes both the knee and the hip. Three functionally separate motoneuron groups are recruited independently to perform one of three tasks: knee extension, knee and hip flexion, or knee extension and hip flexion. Depending on function, the single sartorius muscle has a variety of neurological responses. Similarly, separate motoneuron groups may exist and be recruited for the fish myomere depending on whether the fish is striking at a prey item, initiating an escape response or swimming steadily.

In order to address fully the potential for regionalization in the epaxial muscle mass, information on the innervation of the arm region is needed. Although Westerfield et al. (1986) provide the most detailed description of teleost neuroanatomy relevant to myomeric function, the innervation of the extreme dorsal or ventral axial muscle (the epaxial and hypaxial arm regions, respectively) remains undescribed. This information is essential for a more specific understanding of how epaxial muscle is activated during prey strike and a more general understanding of the potential for complex patterns of myomeric muscle activity.

The possibility of independent neural control pathways for fish myomeric arm regions is suggested by ontogenetic studies of larval zebrafish myomeres. Forty-eight hours after fertilization under standard conditions, all somitic axial muscle blocks, the precursors to the adult central cone regions, have appeared and assumed their characteristic adult chevron shape with cranially pointing, V-shaped myomeres (Kimmel et al. 1995). The arm regions apparently do not form until 3 weeks post fertilization (Van Raamsdonk et al. 1974). Arm region formation occurs long after the sarcomeres of the central axial blocks have become visibly striated and the fish is capable of swimming by axial bending. It is reasonable to hypothesize that such asynchronous development of the arm and cone regions may perhaps be reflected in the innervation patterns of the two regions. Additional investigations into the ontogenetic timing of myomeric arm development relative to prey-strike ability, specifically to cranial lift, may lend insight into the overall function of epaxial muscle during feeding. Deciphering whether (1) formation and functionality of the myomeric arm region are tightly correlated with prey-strike ability or (2) whether feeding kinematics and changes in diet preference are correlated with the degree of arm development both ontogenetically and phylogenetically may help to shed light on the evolution of myomeric shape.

Observations from the present study re-emphasize the possibilities for functional and behavioral versatility within one muscle segment as well as within the epaxial muscle as a whole. During the specific behavior of prey strike, only certain regions of the epaxial muscle are activated. These data demonstrate that epaxial muscle activity is centered in the dorsalmost slab-like arm region of the epaxial muscle (Fig. 1C,D) and can extend as far back as the second dorsal fin. The medial cone portion of the epaxial muscle (Fig. 1C,D) can be completely inactive while the more dorsal arm region is active (Fig. 4).

Although all parts of the fish myomere may be active synchronously during an escape response (Jayne and Lauder, 1995), the myomere itself does not appear to be the obligatory functional unit of the fish axial muscle system. Interestingly, the functional unit is instead a highly context-specific entity that requires redefinition with each new behavioral situation encountered. Past emphasis on the purely undulatory aspects of myomeric function may have obscured additional functional possibilities of the myomere equally as important as locomotion. Interpretations of overall axial myomeric shape and functionality may benefit from investigations of behaviors other than undulatory locomotion.

Most sincere thanks go to P. C. Wainwright, J. Friel, J. Grubrich and K. Ralston, without whose help and training none of this work would have been possible. All EMG material and recording equipment was provided by P. C. Wainwright. D. Burdick and P. Tiffen offered patient and invaluable assistance with statistical analyses. P. C. Wainwright, S. A. Wainwright, W. Kier, S. Etnier and J. Long gave much appreciated advice on drafts of this manuscript. Two anonymous reviewers also offered helpful and constructive comments. Bass were caught by P. C. Wainwright, J. Grubrich, B. Kempter and C. King or purchased from Doc Lewis and C. Terry of the Carolina Fish Hatchery in Faison, NC, USA. This study was funded by NSF grant no. IBN-9306672 to P. C. Wainwright.