Head Waving in Aplysia Californica: II. Functional Anatomy and Muscular Activity During Behaviour

As described in the previous paper (Kuenzi and Carew, 1994a), head-waving behaviour of Aplysia is a searching mechanism that allows the animal to explore its local environment. It is expressed in a variety of contexts, such as foraging, egg laying and locomotion (Preston and Lee, 1973; Cobbs and Pinsker, 1982; Hening et al. 1979). In the laboratory, a bout of head waving lasting 10 min or more can be elicited either by brushing a sprig of seaweed across the animal’s head (Kupfermann, 1974a) or by allowing it to crawl to the edge of a platform (Kuenzi and Carew, 1994a). To initiate head waving, the animal assumes a characteristic posture in which the posterior portion of the foot grips the substratum and the anterior part of the foot and body are lifted off the substratum. The previous paper provided a behavioural description of the bending and twisting of the anterior part of the body that is required to move the head horizontally or vertically with respect to the holdfast (Kuenzi and Carew, 1994a). In the present paper, we describe the muscular system of the body wall that underlies these movements and the coordination of these muscles during head waving.

In biomechanical terms, the body plan of Aplysia is an elongate, fluid-muscle hydrostat; that is, the body consists of a central cavity filled with haemolymph and deformable tissue, and this is enclosed by a muscular body wall. In such systems, connective tissue and muscles in the body wall act in parallel to contain the haemolymph and to establish a positive internal pressure with respect to the environment. In addition, the body-wall muscles serve as the motor system for manoeuvring the body during behaviour. The simplest functional motor system in hydrostatic organisms consists of longitudinal and circular muscles, which provide a mechanism for elongation, shortening and bending of the body (Chapman, 1950; Kier and Smith, 1985; Wadepuhl and Beyn, 1989). Torsion or twisting is effected by activation of helical muscles in the body wall (Kier and Smith, 1985). Although many of the head-waving movements of Aplysia involve body torsion (Kuenzi and Carew, 1994a), previous anatomical studies of the body wall do not report the expected musculature (Brace, 1977a; Eales, 1921). We have therefore re-examined the musculature of the body wall to identify the muscle systems underlying the full range of movements observed during head waving.

Tension in any of the body-wall muscles acts to compress or deform the central fluid space, which redistributes the force to the rest of the body. This space serves as a hydrostatic skeleton in Aplysia, as in other soft-bodied organisms. The pressure in this space is a function of the net activity of body-wall muscles; so, monitoring the hydrostatic pressure during behaviour provides an overall view of muscle coordination. In addition, the electromyographic (EMG) activity of longitudinal muscles can be recorded in intact, freely moving animals (Cook and Carew, 1989; Teyke et al. 1990). Directed head-waving movements, such as those involved in feeding, are associated with a burst of EMG activity in ipsilateral muscles at the beginning of a movement, followed by less intense, but sustained, activity throughout the remainder of the movement. There is no correlated activity in contralateral muscles (Teyke et al. 1990). The movements of undirected head waving during searching behaviour, however, differ from these directed movements in their time course and pattern of body shape changes (Kuenzi and Carew, 1994a). A central goal of this study was therefore to determine the patterns of coordination between muscle groups during head waving. Such an understanding is essential for a complete analysis of the cellular basis of this complex behaviour.

A portion of this study has been published in abstract form (Kuenzi and Carew, 1988).

Aplysia californica were obtained either from the wild, through a commercial supplier (Marinus, Inc., Long Beach, California, USA), or from a laboratory culture maintained at the University of Miami, USA. All animals were held for at least 48 h in a 530 l capacity fibreglass aquarium, filled with recirculating artificial sea water (artificial sea water) cooled to 16°C; they were fed with the seaweed Gracillaria sp.

The macroscopic organization of body-wall muscles was studied by gross dissection of fixed tissue. For fixation, an animal was anaesthetized by injection of 50 ml of isotonic magnesium chloride, and the circumoesophageal ganglia and viscera were removed through a ventral, sagittal incision extending the length of the foot. The animal was then submerged in the anaesthetic and refrigerated for more than 2 h, after which the bathing solution was replaced with 10% buffered formalin in isotonic magnesium chloride. This fixed the tissue without inducing muscle contractions.

Small (4–5 mm) square sections of fixed body wall were further processed for histochemical analysis. The tissue was washed in deionized water, dehydrated in an ethanol series, and embedded in paraplast. Serial sections (10–15 μm) were cut transversely, longitudinally or tangentially, and stained with Mallory’s trichrome (Edwards, 1950). This rendered the connective tissue blue, myocytes red, nerves pink and epidermis reddish brown.

The correspondence between hydrostatic pressure changes and head waving was studied in animals (mass greater than 150 g) isolated from the substratum by being suspended in a tank of sea water. To suspend the animal, a stainless-steel hook attached to a thread was inserted into each parapodium just anterior to its midpoint. The free end of the thread was then attached to a crossbar above the aquarium, as described previously (Cook and Carew, 1989; Kuenzi and Carew, 1994a). Pressure was measured with a Gould DXT disposable pressure transducer (Spectramed, Hawthorne, New York) connected to the animal by a short length of Sylastic tubing (i.d. 1.02 mm, Dow Corning, Midland Michigan) terminating in an 18 gauge hypodermic needle. The whole system was filled with buffered artificial sea water. To insert the needle, the animal was held above the water with its head down, allowing the anterior half of the animal to fill with haemolymph. The needle was then inserted either perpendicular to the surface of the body, into the visceral sinus (haemocoel), or tangentially, into the body-wall haemolymph sinus. The position of the needle’s opening was verified after all recording sessions by post mortem dissection. The transducer output was zeroed using an identical needle and tubing immersed in the tank, and calibrated by measuring the output response to known heights of sea water (1.004 kPa cm⁻¹ sea water). Periodically during a recording session, the system was flushed with 1–2 ml of buffered artificial sea water to remove tissue or small air bubbles that may have plugged the cannula or line. This procedure had no lasting effect on the animal’s hydrostatic pressure. The records of pressure and the animal’s movement were synchronized by split-frame video recording of the front and top (mirror image) views of the animal, and the chart recorder output. During a session, we also observed the hydrostatic pressure in animals that were crawling and head waving while attached to the substratum.

Bipolar cuff electrodes were constructed from two lengths of insulated fine stainless-steel wire (30 μm, Tri-Ml, California Fine Wire), with a 4 mm section of insulation scraped from the tip of each (Fig. 1A). Both were inserted through the wall of a 5 mm section of Silastic tubing (i.d. 5.1 mm, Dow Corning). The tubing was slit along its length, and one electrode was bent to lie flat along the inside wall (active electrode, AE), while the other was bent to project through the slit, at right angles to the tubing (reference electrode, RE). The wires were secured to the tubing and held together along their length with a thin coating of dental impression rubber (Kerr, Romulus, MI, USA). The fine wire leads were longer than the depth of the tank, so that the solder joint between the fine wire and the copper leads could be kept dry. Nevertheless, this joint was thoroughly coated with the same rubber compound for insulation and protection from sea water during the animal’s recovery from surgery.

The electrodes were implanted in animals that had been anaesthetized by injecting isotonic magnesium chloride (50% of body mass) into the haemocoel. A small incision was made through the outer layers of body wall to expose the muscles of the large fascicle layer (see below), and a section of longitudinal muscle was dissected free by cutting away the attached helical and circular muscle fascicles. To implant the cuff electrodes, the whole cuff was inserted under the muscle, and then the slit in the cuff was opened until the exposed section of muscle was completely inside (Fig. 1A). A length of braided silk suture was used to secure each end of the cuff around the muscle. Thus, the final electrode configuration consisted of an active electrode pressed against the surface of the muscle by the cuff and a reference electrode protruding outwards from the cuff. The wire leads extended back through the body wall to the outside. The incision was closed with silk sutures, and an additional suture secured the wire leads to the surface of the body approximately 1 cm from the incision. After the incision had been closed, the animal received an injection of buffered artificial sea water (33% of body mass) and was returned to the holding tank for recovery. Animals began crawling locomotion within 6 h, and behavioural and EMG recording began 24 h after surgery.

Electromyographic (EMG) activity was recorded from large longitudinal muscle fascicles in a layer of the body wall immediately external to a layer of crossed helical fibres (described below). Recordings were made from three locations in the body (Fig. 1B): (a) the anterior, ventrolateral body wall near the rhinophores; (b) the mid-body, dorsolateral body wall immediately anterior to the anterior margin of the parapodia; and (c) the ventrolateral body wall of the posterior neck, near the anterior border of the parapodia. Location c was sampled in every preparation (either on the left or right side), and this was compared with (1) a bilaterally symmetrical site on the contralateral side of the body, (2) location a on the ipsilateral side, or (3) location b on the ipsilateral side. Each combination was observed in at least two animals. The signals were amplified by a.c.-coupled amplifiers (Grass P511, Quincy, MA, USA), filtered by a 60 Hz notch filter and a band-pass filter with cut-off frequencies of 10 and 100 Hz, and recorded on an FM instrumentation recorder (Vetter, Rebersberg, PA) for later replay and analysis. In some cases, the EMG was rectified and integrated to facilitate analysis. After 2–3 videotaped EMG recording sessions, the animal was killed and the position of the active electrode confirmed. The health of the muscle inside the cuff was assessed by comparison with the neighbouring, undissected muscle fibres. Cuffed muscle remained viable and yielded good EMG signals for up to 3 days after implantation. After this, burst activity (see below) was greatly reduced, and the surface of the cuffed muscle became ensheathed with scar tissue.

To synchronize the pressure recordings with measurements of the horizontal and vertical components of the body column/head angle (hereafter referred to as simply the ‘head angle’), an image of the hydrostatic pressure trace on a chart recorder was filmed in the same screen as the dorsal and frontal views of the animal. The position of the animal’s centre of gravity remained stationary, so the horizontal and vertical head angles could be computed by trigonometric transformation from the position of the animal’s mouth in frontal view.

For the EMG portion of this study, head-waving movements were measured as described in the preceding paper (Kuenzi and Carew, 1994a). Briefly, the EMG record and video record were replayed simultaneously and synchronized by their audio tracks. The inner angle between the longitudinal axes of the holdfast and the body column/head region was tracked manually, using a hand-held device that converted the angle into an analogue voltage signal for display on a rectilinear chart recorder in parallel with the replay of the EMG signal. The horizontal and vertical head angles were measured separately, and the records were aligned using the common EMG record. This process was much more efficient and more accurate than the frame-by-frame digitization used to analyze the pressure recordings, and it was sufficient for the qualitative purposes of this study.

In the analysis of anterior versus posterior EMG activity, the head angle was subdivided into a base angle, measured between the longitudinal axes of the holdfast and the body column; and a head angle, measured between the longitudinal axes of the body column and the head.

Head-waving movements are produced by different patterns of activity in the body-wall muscles. These muscles are mechanically coupled both directly, by connective tissue, and indirectly, through the central fluid ‘skeleton’ of internal organs and haemolymph. In such ‘fluid–muscle hydrostats’, activity in a block of muscle not only generates tension parallel to the muscle’s longitudinal axis but, unless it is coordinated with relaxation of other muscle groups, also increases the pressure in the fluid space, applying force to the surface of the entire body wall (Chapman, 1950). Thus, the biomechanical consequences of a muscle’s activity depend on both its orientation, i.e. direction of pull, and its coordination with other muscle groups. The overall coordination of body-wall muscles during head waving can be examined by monitoring the hydrostatic pressure of the haemocoel, and the coordination of specific longitudinal muscles can be monitored by recording their electromyographic activity.

Macroscopic bundles of fibres in the body wall include both muscle and connective tissue, so we used the Mallory trichrome histological technique to differentiate these components. Three general classes of muscle could be distinguished on the basis of both fibre size, i.e. the number of myocytes in cross section and the distribution of connective tissue around both the individual myocytes (endomysium, by analogy to vertebrate muscle) and the bundle itself (epimysium). (1) Muscle bands are large strap-like muscles with a thin connective tissue epimysium and very little endomysium. (2) Large fascicles are smaller than muscle bands, and have thick endo-and epimysia (Fig. 2A,B). (3) Small fascicles contain only hundreds of myocytes in cross section and are richly invested with connective tissue (Fig. 2C).

The body-wall muscles are oriented distinctly as longitudinal, circular and crossed (left-and right-handed) helical muscles (Fig. 2). These are illustrated as they appear in dissection in Fig. 3 and schematically as they would appear in an animal during head waving in Fig. 4. Muscles of different orientation are mechanically coupled at their points of intersection by a continuous epimysia. Two large, longitudinal muscle masses lie on the inner dorsolateral surface and bulge into the haemocoel. These are bundles of muscle bands, with adjacent bands connected so tightly to each other along their edges that they cannot be separated without tearing the muscle. Brace (1977b) named them the lateral columellar muscles for their possible homology with the columellar muscles of more primitive opisthobranch molluscs (Fig. 3A, Lcm). A single layer of longitudinal muscle bands continues across the dorsal midline on the inner surface of the body wall. Immediately dorsal to the Lcms lie a series of longitudinal muscle bands (dorsolateral longitudinal muscles; D-Llm) that are also connected along their edges, but which can be separated easily with a probe. The longitudinal muscles of the dorsal body wall (Dlm) are demarcated by a series of transverse muscles (Dtm) coursing between the longitudinal muscle bands on either side. According to Brace (1977b), both groups of longitudinal muscles originate on the shell and are derived from the intrinsic longitudinal muscles of the body wall (described below). Separate longitudinal muscle bands are also present on the inner surface of the ventral body wall, adjacent to the foot (the longitudinal muscles of the foot, Flm). These originate in the tail, ventral to the Lcms, but their insertion is in the propodium (Brace, 1977b). The inner surface of the posterior, lateral body wall is dominated by a network of dorsoventral muscles (d-vm) and oblique muscles coursing both anteriorly (Aom) and posteriorly (Pom) from their origins on the shell. In this region, the Lcms pass external to these muscles (Brace, 1977b).

External to the above muscle groups, body-wall muscles are distributed more uniformly around the circumference and include circular (c) and left-(lhm) and right-handed (rhm) helical muscles (Figs 2, 3B, 4). These fibres are visible on the inner surface of the body wall ventral to the border of the Lcms, but, on removal of the Lcms, individual fibres can be traced across the midline to the other side of the body. The circular and helical muscles form diffuse inner and outer layers, with bands of longitudinal muscles woven between them. All these muscles are of the large fascicle type, although the helical and circular muscles are significantly thinner and broader than the longitudinal muscles. The fascicles of the outer layer (Fig. 2B) are smaller and more richly invested with connective tissue than those of the inner layer (Fig. 2A).

The next most external layer consists of small muscle fascicles (Fig. 2C). These are continuations of the inner large fascicles, which defasciculate as they course externally to insert on the integument. A network of haemolymph sinuses separates the layers of larger fascicles from these external muscles. Fascicle size is relatively constant, the orientation is evenly distributed among longitudinal, circular and helical (both left-and right-handed) fibres and there is no distinct layering according to orientation, giving this layer a ‘spongy’ appearance in fresh tissue.

In organisms with hydrostatic skeletons, the inner fluid space mediates the mechanical coupling of antagonistic muscles in different regions of the body and so serves the biomechanical role of a skeleton. Monitoring the pressure of this fluid relative to the environment is equivalent to monitoring the stress and compression in rigid skeletons and can provide a measure of antagonist muscle co-activation during movement. In resting animals, the basal pressure in the haemocoel (visceral sinus) was 0.19±0.019 kPa (S.E.M., N=23 preparations). This is relatively low compared with that in terrestrial gastropods (for a review, see Chapman, 1975), but is similar to that of other marine gastropods, e.g. Bullia sp., 0.15 kPa (Trueman and Brown, 1976). In active Aplysia, the highest transient pressure observed was 0.75 kPa. The basal pressure is likely to be due to tonic contraction of the body-wall muscles alone. This was demonstrated by injecting an anaesthetised animal with a large volume of isotonic magnesium chloride (equal to the mass of the animal and nearly doubling its volume). In spite of the large volume change, the basal pressure fell to zero relative to the pressure of the surrounding water.

The correlation between pressure transients and head movements in suspended animals was investigated in synchronized video and pressure records during periods of large-amplitude body movements and/or large fluctuations in pressure. The orientation of the anterior part of the body with respect to the posterior part was resolved into horizontal and vertical components.

Purely horizontal head movements were not associated with pressure transients. In the example shown in Fig. 5, the animal made marked lateral bending movements, indicated by hatched bars, yet there were no consistent and measurable changes in pressure. Similar results were observed in the three animals for which kinematic data are available, and in 24 other preparations, including both suspended and substratum-attached animals. This indicates that during these bending movements muscle contraction on one side of the body is coupled with relaxation, or a decrease in activity, of other muscles. Thus, the internal pressure does not change. In addition, because the body does not shorten during head-waving movements, the activity of circular muscles should be constant, suggesting that bilateral longitudinal muscles are coordinated in antiphase during lateral movements. This was tested directly using EMG recording (see below).

Vertical movements, however, were consistently associated with pressure transients in suspended animals. For example, in Fig. 5, the animal lifted its head three times (open bars). These movements corresponded to three pressure pulses (peaks marked by arrows in both traces). These pulses were large relative to both the basal pressure (here 0.12 kPa) and the surrounding pressure fluctuations. There was no difference between the transients that occurred during purely vertical movements (first pulse) and those that occurred during a diagonal sweep (movement with horizontal and vertical components indicated by the stippled bar). Although more sustained head lifts corresponded to more prolonged increases in pressure, the duration of the pressure pulses was shorter than the duration of the movement, and the onset of a pressure pulse was delayed relative to the beginning of the corresponding head lift. Therefore, the animal did not appear to be using the energy stored in the pressure of the haemocoel to effect the lifting movement, but the pressure pulse was probably an effect of muscle contractions during the lift. Such pressure changes were not observed when animals attached to the substratum made similar movements. The results from 24 suspended animal preparations were qualitatively similar, in that pressure transients only occurred in conjunction with movements that included a marked vertical component, i.e. head lifting or diagonal sweeping, and were not observed during purely horizontal movements.

A final question of movement biomechanics in hydrostatic organisms is whether pressure in different compartments can be regulated independently. Three such compartments might exist in Aplysia: the haemocoel and the left and right body-wall haemolymph sinuses. The independence of these spaces was tested by simultaneously recording the pressure in two potential compartments. Only minor differences in basal pressure and the relative amplitude of pressure transients were observed (Fig. 6). Specifically, the basal pressure of the body-wall sinus was 0.164±0.025 kPa (S.E.M., N=13), and in preparations where both spaces were monitored, the pressure difference (haemocoel minus body wall) was -0.018±0.019 kPa (P>0.05, two-tailed t-test, N=12).

The spatial and temporal patterns of muscle activation in the body wall of Aplysia can be observed by electromyographic (EMG) recording. For this we chose to record from the large fascicle longitudinal muscles, as they provide the most mechanically stable and electrically isolated substratum for cuff electrode recording and they are represented throughout the body. Helical and circular muscles proved to be too delicate for the surgery and handling necessary for the preparation. During head waving, EMG activity in these muscles was characterized by bursts of spikes (Figs 7–9). The individual spikes were triphasic and slow compared with extracellularly recorded nerve action potentials (200 ms for EMG, 20 ms for nerve), and in this regard they were exactly equivalent to the signals recorded from the lcms, as reported by Cook and Carew (1989). Although some units could be recognized throughout a recording session by their relatively constant amplitude, the amplitude of others appeared to increase and then decrease during a burst. The wave-form of each of these latter spikes was also triphasic and did not, therefore, appear to be a complex summation of smaller units. In general, the number and frequency of spikes in a burst tended to vary with the duration and velocity of the turning movement. The periodicity of the bursts varied with the period of horizontal bending in these animals, which ranged from 15 to 30 s.

Activity of the ventrolateral muscles was closely coupled with horizontal movements towards the side of the body containing the recording electrode (Fig. 7). The onset of a burst preceded the beginning of movement by 2–3 s, and EMG activity continued throughout the movement, although the greatest activity was during the first 10 s of movement. Larger-amplitude movements often consisted of two bursts separated by a pause, with each burst being 5–10 s in duration. As reported above, the hydrostatic pressure of the haemocoel did not vary systematically with horizontal movements, suggesting that bursts of activity on opposite sides of the body should not overlap. This was consistently observed, even when the animal switched rapidly between bending towards different sides, as shown at the end of the trace in Fig. 7. In general, activity on one side terminated, or decreased markedly, at the onset of either activity in the contralateral muscle or movement to the contralateral side.

Repetitive movements were not observed while recording EMG activity, so it was difficult to quantify the correspondence between the direction of movement and the magnitude of the EMG signal. Qualitatively, however, muscle activity was correlated with movements that had a marked horizontal component, but was not consistently correlated with vertical movements. In this experiment, the recording sites were slightly ventral to the transverse axis of the animal, at approximately 120° and 240° in circumferential position, where 0° is dorsal. The recorded signals would therefore be expected to be less intense during helical sweeping and upright pivoting than during lateral sweeping of comparable amplitude. In the series shown (in Fig. 7), and in other sequences from this and other animals, there was no systematic trend, and horizontal movements were associated with bursts of similar intensity regardless of the vertical head angle.

Vertical head movements, however, were not consistently accompanied by burst activity at these sites. Again, given the more ventral position of the electrodes, these muscles would be expected to show greater activity during head depression than during head lifting. However, the only systematic correlation was a decrease in activity on one or both sides, or relaxation of the muscle, with head depression (e.g. the movement indicated by the vertical dashed line in Fig. 7).

Longitudinal muscles are distributed around the entire circumference of the body, and there are no physical constraints on the direction in which an animal can bend its body. The simplest model of movement, then, would be that the activity of a longitudinal muscle at a given circumferential position in the body wall would be maximal when the body bent towards that position. For example, a muscle located at 60° (where 0° is dorsal) in the dorsolateral body wall would be maximally active during diagonal sweep movements towards 60°. In addition, muscles on the dorsolateral side of the animal should be more active during the helical sweeping and upright pivoting types of horizontal movements, when the body is more elevated. This was tested by comparing EMG activity in muscles at similar positions along the anterior–posterior axis, but at different circumferential positions. The bilateral recordings shown in Fig. 7 are the extreme of such pair-wise comparisons and they faithfully represent the horizontal movements. In addition, we recorded activity in muscles on the dorsal and ventral sides of the body (e.g. sites b and c in Fig. 8), to test the correlation between their activity and vertical movement and/or position. The longitudinal muscle in the dorsal electrode was part of the muscle band layer, as there are no large fascicle longitudinal muscles in this region.

As before, the ventral muscle recording included several large units whose activity was correlated with horizontal movement towards the ipsilateral side (Fig. 8). However, records from the dorsal muscle in this and two other preparations included many small units that were tonically active, so modulation of activity in the dorsal electrode is shown more clearly in the integrated records in the lower traces of Fig. 8. Bursts of spikes in the more dorsal muscle were correlated with both head lifting and horizontal bending, especially when the latter movements were helical sweeps or upright pivots to the ipsilateral side; that is, when the body was erect during the movement (behavioural categories according to Kuenzi and Carew, 1994a). During the long lateral sweep indicated by the arrow in Fig. 8, the dorsal muscle was relatively quiet while the vertical angle was low, but it began a burst during the lifting movement. In contrast, the ventral muscle fired for the duration of the movement. Depression of the head was accompanied by relaxation at both sites (Fig. 8, arrowheads). The orientation of the animal would have allowed this movement to be passive (assisted by gravity), or it may have been driven by foot muscles that were not sampled.

Differences in the activation of anterior and posterior regions of the same muscle depended on the animal’s degree of extension. When animals head-waved while relatively contracted, i.e. while the anterior margin of the holdfast was near the anterior electrode (site a), there was no discernible difference between activity in the two electrode positions (sites a and c). As the animal extended, the amount of activity in the anterior electrode and its modulation with head-waving movements were reduced. In contrast, activity in the posterior electrode was modulated as observed above, with bursts of activity corresponding to ipsilateral bending, for example (Fig. 9). Units with large signals in the anterior recording were active while the animal crawled on the platform (Fig. 9B), indicating that the muscle at this site was still healthy, but there was little activity during head waving when the body was in a more extended position (Fig. 9A).

The relative lack of activity in the anterior muscle was unexpected because this site is in a region of significant local bending when the animal is extended. To examine this in more detail, the horizontal angle was subdivided into base and head angles as described in the Materials and methods section. These measurements are shown in the middle traces of Fig. 9A. In this example, the base of the animal was bent towards the right (ipsilateral to the electrodes) throughout the time shown (dashed line). However, the animal bent its head further to the right several times before bending it to the left near the end of the trace. Each bend to the right was accompanied by bursts of activity in the posterior muscle, whereas activity in the anterior muscle increased noticeably only in the most extreme bends. In both cases, activity decreased sharply when the head returned to the straight position with respect to the body column, and bending to the left (contralateral) interrupted, or greatly reduced, activity in both muscles.

Activity in the anterior muscle increased greatly when the animal tucked and began crawling on the platform (open bar in Fig. 9B). In contrast, during the time indicated by the stippled bar the animal made a helical sweep movement to the right, which was accompanied by bursts of activity in the posterior electrode.

The difference in muscle activity therefore depended on the relative location of the holdfast, suggesting that most of the torque for head waving was generated posteriorly. For most of the 3 h of records (from two animals), the position of the posterior recording site was anterior to the holdfast, in the body column region. In this sample, the results are consistent with a decreasing gradient of activity extending anteriorly from the holdfast, where the modulation of activity is maximal.

The body-wall muscular system of animals with hydrostatic skeletons is characterized by an orthogonal array of longitudinal and circular (or transverse) muscles (Chapman, 1950). In addition, an array of left-and right-handed (crossed) helical muscles is necessary to provide the mechanical action for torsion of the body (Kier, 1982; Kier and Smith, 1985). In Aplysia, all of these elements are present in the layers of intrinsic body-wall muscles. The most internal longitudinal muscles are especially distinctive. These include the lateral columellar muscles and the dorsal muscle bands, which form a nearly continuous sheet of longitudinal muscle over the dorsal hemi-cylinder of the animal. The inner longitudinal muscles in the ventral portion of the animal are also distinctive, but consist of separate muscle bands interwoven with helical and circular muscle. The other major muscles of the body wall are arranged according to the generalized hydrostatic body plan, including circular and left-and right-handed helical fibres (Wainwright et al. 1976). The body wall of the pulmonate mollusc Lymnaea stagnalis is also arranged according to this plan although, in this species, the orientation of more posterior muscles is strongly affected by the presence of the shell (Plesch et al. 1975).

The columellar muscle of some shelled molluscs can function as a muscular hydrostat, in that the longitudinal bundles are enclosed by layers of circular and helical muscles with no intervening haemocoel (Brown and Trueman, 1982; Trueman and Brown, 1976, 1985). We found no evidence for such a structure in the body wall of Aplysia.

The hydrostatic body plan allows an almost limitless range of body movements, but at a neural level these can be understood on the basis of a few simple rules of muscle coordination. The most simple hydrostatic body plan consists of a central constant-volume cavity, filled with fluid or deformable tissue, enclosed by an epithelium and a network of circular and longitudinal muscles. Antiphasic activation of these two groups of muscles is sufficient to produce elongation and shortening movements. This mechanism is used, for example, by worms burrowing through a substratum (Chapman, 1950). Bending movements are added to this framework simply by providing separate neural control of the longitudinal muscles according to their position around the circumference of the body, e.g. dorsal and ventral or left and right groups, and maintaining the activity of circular muscles to resist an increase in diameter resulting from shortening of one side of the body (Kier and Smith, 1985). Further activation of circular muscles amplifies the effect of longitudinal muscle activation (Wadepuhl and Beyn, 1989). Thus, an effector system consisting of only two anatomically distinct muscle types, longitudinal and circular, can be differentially organized at the neural level to produce two separate, complex patterns of movement: (i) elongation and shortening for forward locomotion; and (ii) bending. Adding muscles wound around the body in left- and right-handed helices provides a third type of movement; torsion of the body, or rotation about the longitudinal axis (Kier, 1982; Kier and Smith, 1985). In the squid, this type of motion is used to align the ends of the tentacles with the prey during a predatory strike (Kier, 1982). In Aplysia, we have found that torsion stabilizes the motion of the head while the body bends horizontally during head waving (Kuenzi and Carew, 1994a). The hydrostatic skeleton mechanically links all muscles in the body wall, so the nervous system must coordinate activation and relaxation of all the different groups in order to produce any type of movement.

The finding that there are no pressure changes during side-to-side bending allows us to make two predictions concerning the possible patterns of muscle activation for this kind of movement. First, the timing of activation of muscles on opposite sides of the body should be antagonistic, with little or no co-contraction during body bending. Second, once the animal has bent to one side, movement in the opposite direction requires muscle activity on the side of subsequent bending, i.e. there is no passive return to centre. In suspended animals, the observation that pressure pulses were delayed relative to vertical bending movements suggests that, although the initial movement may be due to simple longitudinal muscle contraction, other muscle groups are recruited to sustain the bend, perhaps to counteract the force of gravity or to redistribute haemolymph. Alternatively, the pulses may be the result of contractions of the tail, as might be used in gripping the substratum when one is available.

The EMG records from longitudinal muscles in different regions of the body wall support these hypotheses. In short, the EMG burst begins just before the start of movement to the ipsilateral side, regardless of the current position of the animal’s head.

For example, the EMG activity in a left lateral muscle is similar if the animal is bending to the left after pausing in the centre, or after it has completed a bend to the right. Therefore, one apparent rule of head-waving coordination is that lateral bending towards one side is accompanied by a burst of EMG activity on that side, regardless of the initial position of the head, i.e. fully flexed to the contralateral side, near the centre or nearly fully flexed to the ipsilateral side. Furthermore, this activity is coupled with a significant decrease in the activity on the contralateral side; again, regardless of initial head position. The muscle activity therefore corresponds to the direction of movement rather than to head position. If head position is compared with the difference in timing between muscle activity on the left and right sides, there is an apparent lag between the two measures, as was also observed by Cook and Carew (1989).

Head lifting and depression also involve bending of the hydrostatic body cylinder, so the underlying muscle activity should be similar to that driving horizontal bending. The helical sweep and upright pivot movements include a tonic elevation of the body, so muscles on the dorsal side of the animal should also be active during these movements. The EMG records from dorsolateral muscles confirmed these predictions for dorsal movements. In contrast, most ventrally directed movements were not correlated with activity in ventrolateral muscles. In this study, ventral movements would be assisted by gravity, so they may not require the same changes in activity as the corresponding dorsally directed movements, or they may be mediated by foot muscles. However, the sharp ventral flexion described as ‘tucking’ (Kuenzi and Carew, 1994a) was accompanied by a large, long burst of EMG activity in the ventrolateral muscles, suggesting that this movement of the body is active.

The topography of head-waving movements suggests two loci of muscular activity: (1) the junction of the head and body column regions; and (2) the junction of the body column and holdfast (Kuenzi and Carew, 1994a). The EMG records, however, support only one site of activity, near the holdfast. The anterior bending site may result from differences in the stiffness of the body along the longitudinal axis. Specifically, the head region contains the relatively stiff buccal mass, whereas the body column region contains the crop, gizzard and hepatopancreas. The relatively flexible oesophagus joins the buccal mass to the crop and lies at the junction of the head and body column (Kandel, 1979). Thus, the mechanical properties of the internal organs provide a substratum for the observed local bending of the head without necessitating local muscular activity at the head/body column junction.

The initiation and coordination of muscle activity is mediated by neurones in the circumoesophageal ganglia. Lesions of interganglionic pathways between circumoesophageal ganglia significantly alter or impair coordinated movements such as crawling locomotion (Jahan-Parwar and Fredman, 1979), head waving (Kupfermann, 1974b; Kuenzi and Carew, 1994b) and egg laying (Ferguson et al. 1989). In addition, if segments of body wall are physically dissociated except for their connection to the circumoesophageal ganglia through separate peripheral nerves, responses to stretch of one segment are observed in other segments (Jahan-Parwar and Fredman, 1978).

Identification of patterns of motor coordination in muscles should, therefore, provide criteria for assessing the role of central neurones in production and coordination of the behaviour as well as providing signposts for developing more reduced preparations. This study suggests three such patterns.

Muscle activation on one side occurs simultaneously with a sharp decrease in activity, or cessation of activity, on the other side. There are two possible neural pathways for coordinating this antagonistic, bilateral interaction: the pedal commissure and the cerebral commissure. Several studies on locomotion in Aplysia, either swimming (in species that swim) or crawling (in benthic species), have shown that the bilateral synchrony of these behaviours can be disrupted by transection of the pedal commissure (Parsons and Pinsker, 1988, 1989; von der Porten et al. 1980; Kupfermann, 1974b). Whichever pathway is responsible for the bilateral coordination, the cellular circuit will include interganglionic interneurones, as there are no pathways for direct bilateral interaction between efferent neurones. The difficulty thus far in recording from circular muscles prevents formulation of any prediction concerning activity patterns of motor neurones that drive circular muscles. However, on the basis of the biomechanics of hydrostatic systems, the circular muscle activity should either not change, or increase, with large-amplitude bending (Wadepuhl and Beyn, 1989).

Even though the freely moving animal’s movements appear to be complicated by the independent movements of different body segments, these separate movements are not, surprisingly, the result of local muscular activity. Thus, movements of large portions of the animal’s body are the result of activation of only a small portion of the posterior body muscles. Motor neurones in the pedal ganglia innervate distinct regions of the body wall (Hening et al. 1979). We predict, therefore, that the motor neurones innervating the more posterior body-wall muscle should show the greatest modulation during head waving.

Undirected head waving is a sequence of short (4–10 s), relatively discrete movements directed either horizontally or vertically. Longer-duration movements are composed of several bursts of movement in the same direction; these bursts also having a period of 4–10 s (Kuenzi and Carew, 1994a). This behaviour is reflected in the EMG recordings as alternating bursts of electrical activity in the left and right longitudinal muscles during rapid side-to-side bending and a longer series of bursts for longer movements in the same direction. These bursts continue throughout the movement.

Body-wall muscles in Aplysia are innervated directly by motor neurones (McPherson and Blankenship, 1991), so the burst pattern in EMG records is expected to resemble closely the spike pattern in motor neurones. The motor neurones innervating the body-wall muscle are located primarily in the ipsilateral pedal ganglion, and lesion studies suggest that the oscillator circuits for both pedal locomotion and swimming are also located, at least in part, in the pedal ganglia (Bablanian et al. 1987; Hening et al. 1979; von der Porten et al. 1980). Our behavioural studies suggest that undirected head waving is also influenced by a circuit capable of producing rhythmic bursts of movement; the output of this circuit must be coordinated within the central nervous system to give rise to the overall pattern of complex motor output (Kuenzi and Carew, 1994a). The results of both the present paper and the preceding one suggest that the muscle activity responsible for each component of head waving is coordinated with the activity of other body-wall muscles (1) along the longitudinal axis, (2) bilaterally and (3) between muscles of different orientation. Important aspects of these interactions depend on interganglionic communication. In the final paper of this series (Kuenzi and Carew, 1994b), we explore the roles of the principal interganglionic pathways in controlling and coordinating head-waving movements.

This work was supported by NIH training grant HD07180-07, a grant from the Connecticut Department of Higher Education to F.M.K. and NIH grant RO1-MH-14-1083 to T.J.C.