ABSTRACT
Methods are described for suspending and clamping Aplysia fasciata so as to permit intrasomatic recording from neurones of the head ganglia during locomotor and other behavioural activities.
Sensory responses of neurones in the pedal ganglion are classified into four main types, all being distinct from those of pleural ganglion cells.
The pedal ganglion may well contain ‘motor cells’ for the greater part of the somatic musculature.
Preliminary results suggest that the pleural LGC may be involved in promoting a change from swimming to creeping behaviour.
INTRODUCTION
By reason of their large size and ease of penetration with single or multiple microelectrodes, the central neurones of opisthobranch and pulmonate gastropods have contributed a substantial proportion of current knowledge of invertebrate synaptic mechanisms, integrative processes and neurone geometry (Tauc, 1966). By contrast, far less is known of the behavioural roles of these same neurones, and in consequence it is difficult to assess the significance of the various postulated integrative mechanisms (Frazier et al. 1967).
The work of Willows (1968) and Willows & Hoyle (1968) was an important contribution in that for the first time behavioural acts were evoked from a gastropod by direct stimulation of certain specific single cells. The most dramatic of these acts was identical with the normal evasive activity of Tritonia when stimulated by the proximity of a starfish normally preying upon it. In addition to providing information about the functional significance of patterns of neural circuitry, the Tritonia preparation indicated the danger of assuming that isolated parts of nervous systems behave in the same way as when they are connected to their sensory inputs or to other ganglia. The majority of cells which would have exhibited pacemaking in isolated brains were silent unless stimulated when their peripheral connexions were intact (Willows, 1968). A similar phenomenon was reported more recently by Stinnakre & Tauc (1969) in Aplysia abdominal ganglia, where the well-known pacemaking cell R. 15 (terminology of Frazier et al. 1967) only becomes regularly autoactive once its connexions with the branchial ganglion and/or osphradium have been severed. Pacemaking in this case is normally inhibited by the action of hypo-osmotic sea water.
Two more recent studies (Kupferman & Kandel, 1969; Peretz, 1969) described various features of an abdominal ganglion preparation in Aplysia californica where connexions with the gill and organs of the mantle shelf were left intact. As in Tritonia, stimulation of certain motor cells evoked stereotyped and repeatable movements, whereas others evoked more complex effects indirectly via these ‘motor neurones’. The behavioural functions of these movements were presumed to relate to gill ventilation, expulsion of faeces and detritus or to protective withdrawal of the delicate respiratory exchange structures.
Thus the majority of gastropod behaviour patterns studied with intra-cellular microelectrode techniques have so far been concerned with escape or evasion-related activities: retraction of branchial tufts, turning and swimming in Tritonia; gill withdrawal in Aplysia. It is perhaps not unexpected that this should be so, for the condition of most neurophysiological experiments must be fairly noxious from the point of view of the animal. Though there are undoubted technical advantages in working with escape-mechanisms (large size of cells involved, ease of evocation of behaviour), it is probable that study of these alone would yield an unduly simplified view of the neural mechanisms underlying behaviour. For instance, Dorsett, Willows & Hoyle (1969) reported that isolated Tritonia brains will generate patterns of neural activity indistinguishable from evasive swimming when triggered by electrical stimulation of either the second or third cerebral nerve. Such independence from sensory feedback would not be expected in the less ‘all-or-nothing’ movements of normal locomotion in view of the known existence of proprioceptive reflexes in other gastropods (Herter, 1931a, b).
One of the large European species of tectibranch, Aplysia fasciata, is a powerful swimmer capable of covering considerable distances by active linear progression. Unlike the swimming of Tritonia, this is not an escape activity. The reaction of this species of Aplysia to noxious stimuli is to contract into a ball and to emit purple ink and mucous secretions. The three components of the response are recruited progressively according to the intensity of the stimulus. Swimming, on the other hand, can reliably be evoked in the intact animal by suspending it free of the substratum. The present work was initiated to develop a preparation which would permit intraso-matic recording from central neurones during behavioural activities, especially the swimming referred to above.
METHODS
It was found comparatively easy to evoke swimming from ‘tethered’ intact Aplysia, even when they were quite firmly restrained by eight hooks (Fig. 1 ; Plate 1). The experimental tank was large enough to accommodate animals up to 900 g suspended in this way, and the depth such that the fully turgid foot just failed to touch the bottom while the visceral hump and siphon were fully immersed. The sea water was continuously renewed at 250 ml to one litre per min from the bench supply line, temperature controlled to 17 +1 degC and aerated. Text-fig. 1 shows the response of a swimming undissected animal suspended in this way to objects touching the foot-very reminiscent of the flying insect’s tarsal inhibitory reflex.
Unfortunately, it was found that as soon as the animal was opened to expose the anterior ganglia, the greater proportion of the haemolymph rapidly escaped. In Tri-tonia the condensed head ganglia allow the buccal mass to be used as an internal plug to minimize this (Dorsett, personal communication). Such bled animals, though they will perform repeated parapodial folding and other movements for about 30 min, soon become inert except for single closure movements of the branchial chamber and parapodia every 2–4 min (the SGM of Peretz, 1969). Various methods were tried for success in retaining turgor, but this proved difficult because in order to obtain swimming it was evidently necessary that the foot should not be too strongly stimulated.
The method eventually adopted is shown in Text-fig. 2. First, the animal was suspended dorsal surface uppermost by the eight hooks attached as described under Text-fig. 1. A padded ring consisting of two semi-elliptical bands of Perspex hinged together at one side was then clipped around the ‘neck’ of the animal just anterior to the parapodia. The haemolymph in the posterior part of the body was retained by means of a latex balloon inserted into the haemocoele through a short slit in the body wall just anterior to the clamp ; the balloon was inflated with air once it was in position so that it pressed the body wall outwards against the padded ring, displacing forward the contents of the crop which had previously occupied this position. The balloon consisted of a short length of latex tubing tied on to 3 mm O.D. thick-walled Polythene tubing bent into a right angle which gave the stiffness necessary for its insertion. This operation entailed little loss of haemolymph from the posterior part of the body. However, animals undergoing dissection always contracted the parapodia, thus forcing a considerable proportion of the haemolymph into the neck region ; this escaped during subsequent dissection. In order to re-expand the posterior sinuses to their original volume, a smalldiameter Polythene tube (shown at the top left in Text-fig. 2) was pushed down between the balloon and the body wall. Sea water injected through this seemed to form an adequate substitute for haemolymph.
When this technique was first tried on an animal which had lost most of its haemolymph the effects were dramatic. As the posterior sinuses were expanded, the previously inert animal started waving its parapodia, expanding and contracting its foot and crawling when a suitable substratum was presented. In addition to this seeming tonic excitatory effect of distension, such movements as were occurring became much easier to interpret. The action of the hydrostatic skeleton meant that contraction of, say, parapodial levator muscles resulted in the normal levation movement rather than local shortening which produced nothing more than enigmatic wrinkles.
The dorsal slit was then extended anteriorly, as shown in Text-fig. 2. A pair of hooks were used to open up the slit permitting further dissection. The crop was ligated posteriorly near to the clamp and removed. This allowed the buccal mass to be pulled forward and also exposed the circumoesophageal ring of ganglia, the pedal nerves and the pleurovisceral connectives. A further marked improvement in the activity of the animal was found to result from perfusing the anterior aorta with oxygenated sea water. This was done with a micromanipulated hypodermic needle inserted into the aorta. Animals often swam at this stage and had to be quieted by raising a platform to make contact with the foot (see Text-fig. 1). This platform consisted of a continuous Polythene belt stretched over rollers, which apparently constituted a suitable surface for creeping.
The final step in preparing the animal for microelectrode recording was to insert the ganglion table shown in Text-fig. 3(b) through the circumoesophageal nerve ring bad under the left pedal and pleural ganglia, whose neurones were the subjects of the present investigation. The left side was chosen because the penis obscures the nerves on the right. Micropins through the outer sheaths of these ganglia immobilized and tensioned the surface of the sheath. In early experiments the sheath was slit with a sliver of razor blade; but in addition to damaging a proportion of cells, this procedure ruptured the blood sinuses around the brain so that perfusion of the ganglia became less effective. It may be also that the rather high perfusion pressure normally used (60 cm sea water) distended the anterior foot sinuses with a beneficial effect comparable to that of the balloon on the posterior sinuses, and that de-sheathing reduced the effective hydrostatic action on the foot.
It was found that specially stout microelectrodes mounted on the modified loudspeaker shown in Text-fig. 3 (a) would penetrate all but the thickest sheaths without slitting. The crucial features aiding penetration were the raised high-frequency response of the loudspeaker, the accurately axial movement conferred by the pair of copper diaphragms constituting the coil suspension and the stiffness of the microelectrodes themselves. These electrodes were pulled from thick-walled Pyrex capillary tubing with a modified Nastuk electrode puller which blew cooling jets of air on to the electrode tip at the time of the strong pull Chowdhury (1969). This resulted in electrodes which for the same tip size and shape had a much shorter and stouter shank. Penetration in most cases was best achieved with rectangular current pulses of 1–2 ms duration which produced a withdrawing movement of 30–60 μm amplitude followed by some undamped ‘ringing’.
Amplifying, recording and electronic stimulating equipment was of the conventional type. A Tektronix 564 storage oscilloscope was found very useful because of the long time-scale of most behavioural activities. Behaviour and neurone discharge were viewed in 2 min episodes-longer than the longest persistence phosphor otherwise available. Even longer-term recordings were stored on magnetic tape using a four-channel Thermionics T 3000 FM tape recorder, and relevant sections of such stored records were selected for filming later. Artists’ paint brushes of various sizes were used to deliver tactile stimuli. Extracellular recordings from nerves and connectives were taken with suction electrodes using a multi-way switch to connect recording or stimulating leads to the desired channel. Movements were recorded with one or two light isotonic photoelectric transducers capable of linear registration of movements from 100 μm. to 10 cm. These were routinely attached to the anterior rims of either or both parapodia with hooked micropins (see Plate 1). The loading could be adjusted from 10 to 100 mg according to the desired frequency response. The lightest load gave a high-frequency response of 1 s for the full 10 cm amplitude of movement. The tension in the foot muscles was monitored continuously with a Grass strain gauge whose compliance was varied as desired with springs attached in series or in parallel with the transducer.
RESULTS
Intracellular recording of responses to physiological stimulation
Pedal ganglion
One disadvantage of working with pedal neurones is that, with the exception of about six large cells situated posteromedially on the dorsal surface of the ganglion, these are too numerous and too similar in size and pigmentation to be individually recognizable. Nevertheless it was thought important to investigate their behaviour as far as possible using only an approximate method of grid location and such physiological classifying criteria as presented themselves, since the large and identifiable cells hitherto studied in other ganglia may not be typical in all respects and certainly do not perform all possible integrative tasks. Moreover, the majority of the nerves to the foot and parapodia come from the pedal ganglia, and it is highly probable that the neurones of these large ganglia play an important part in co-ordinating locomotion and other behavioural activities. With these considerations in mind, all cells penetrated were routinely tested for receptive field, axon distribution, firing pattern during spontaneous movements and motor consequences of stimulating the cell directly through the recording electrode.
Text-fig. 4 shows a selection of records taken from various pedal neurones which illustrate different types of sensory response. Out of a sampled population of 180 cells, 14% responded to tactile stimulation of small areas of body surface with intense excitation, bursts of spikes, with or without clear synaptic potentials (Text-fig. 4, type (a); Text-fig. 5). Many of these were normally silent and relatively insensitive to depolarizing current passed through the recording electrode with a bridge network-the latter perhaps because the pacemaker region was electrically remote from the soma. Direct stimulation of such cells evoked no motor effects and their behaviour was what would be expected of first-order sensory interneurones. These ‘type a’ cells resemble the ‘type 1’ cerebral neurones of Anderson (1967) and were the only ones in the present study to show any suggestion of systematic localization in the ganglion with respect to either sensory or motor field or adequate sensory modality. A large proportion of them had receptive fields in the anterior foot or head regions and formed a loosely defined group near to the root of the anterior pedal nerve (see Text-fig. 2).
A rather small proportion of cells showed only inhibitory responses. Text-fig. (a) is an example of this class which constituted only 10% of the total sampled. Receptive fields were mostly large.
Excitatory responses with or without after-inhibition were commoner-33 % of the total (see Text-fig. 4c, d). These were quite distinct from type a in always exhibiting a tonic discharge before stimulation, in their sensitivity to depolarizing current which always evoked a tonic discharge, and most of all in the possession of large and often ill-defined receptive fields ; excitation was usually only moderate in degree. Adequate stimuli for types c and d were also often non-specific with respect to modality. The neurone illustrated in Text-fig. 4(c) responded moderately or weakly to a variety of tactile, proprioceptive-like, vibrational, water-current and diffuse photic stimuli. It is possible that such neurones are only driven strongly by certain precise stimulus configurations which were not hit upon in this survey.
Numerically the largest class (41% of penetrations) were of the type shown in Text-fig. 4(e)-mutually antagonistic receptive fields in different parts of the body. The antagonistic fields tended to be anterior and posterior rather than in bilaterally symmetrical positions. Only 4% of cells failed to respond to any inputs tested.
The size of receptive field in all the above classes varied continuously in extent, but those of type a cells were always smaller than the smallest field cells of other classes, and type a cells never had demonstrable peripheral axons. The criteria separating the other types are less objective and the classification there may only be one of descriptive convenience.
Pleural ganglion
The responses of the large cells in the pleural ganglion were generally more complex, in fact more like the half dozen or so large cells in the (dorsal) posteromedial quadrant of the pedal ganglion referred to above. Most had two-branched or three-branched ipsilateral and/or contralateral peripheral axons and responded phasically to injected current. Waning or facilitation of responses to repeated tactile or other physiological stimuli and the frequent occurrence of after-inhibition or after-excitation were prominent features. The few small pleural neurones which were penetrated had sensory responses more like the pedal cell types b-e described above. In both ganglia the large cells were more heavily pigmented, emphasizing their distinctness. An example of a large pleural neurone which integrated information arriving along several sensory channels is shown in Text-fig. 6. Tactile stimulation of small sensory areas evoked mixed e.p.s.p. and i.p.s.p. during the stimulus, with rebound excitation afterwards. Qualitatively similar stimuli delivered to progressively larger areas evoked progressively more intense ‘on’ bursts. This spatial summation results in a comparatively sophisticated qualitative discrimination of different types of object touching the visceral hump-the receptive field of this neurone. The occurrence of different sizes of incompletely invading ‘A spike’ suggests the existence of spatially separated and independently pace-making integrative loci, possibly on different branched axons as suggested by Tauc & Hughes (1963). This cell had two ipsilateral peripheral axons (Text-fig. 6b). Several pedal type a cells (Text-figs. 4, 5) converging on this pleural cell would be a simple arrangement capable of mediating such spatial summation.
Pedal ganglion ‘motor cells’
In the course of this survey it was found that about 15 % of the cells sampled in the pedal ganglion evoked a reproducible local muscular contraction (see Text-fig. 7). Such cells were observed to fire whenever the same muscles contracted spontaneously and they also had a single unbranched peripheral axon running (ipsilaterally) in the nerve to the activated region. They have in fact most of the characteristics which would be expected of motor neurones and presumably correspond to the cell type b of Willows (1968). Unlike the abdominal ‘motor neurones’ of Kupferman & Kandel (1969), the movements evoked by direct stimulation of this type of pedal cell never, resulted in twitches corresponding one-for-one with spikes in the soma. In view of the existence of a peripheral neural plexus whose function is not known (Bullock & Horridge, 1965), the present data do not prove that these pedal ganglion neurones are the final common path through which all efferent signals must pass; and for this reason they will be termed ‘motor cells’ rather than motoneurones. Nevertheless it seems clear that any further peripheral modification must be slight since the evoked movement bears a close and consistent relation to the frequency at which the cell discharges (see graph Text-fig. 7).
Text-fig. 8 illustrates the properties of a parapodial levator motor cell. Again the movement evoked was smooth and its minimum latency of 400 ms is very long. Assuming that no drop in conduction velocity occurs peripherally due to tapering of the axon, conduction time only accounts for 100 ms. But unlike the gill preparation of Kupferman & Kandel (1969), the transducer in the present experiments could not be attached to the contracting structure and the interposition of extremely compliant viscoelastic tissues in series could well account for the rest of the delay and would also smooth and ‘blunt’ individual twitches if present.
Behavioural activities and patterns of neuronal activity
Creeping
The consistent correlation between spontaneous discharge pattern of such motor cells and behavioural activities occurring at the time is particularly clear in Text-fig. 9. All four of the cells illustrated fired a burst of impulses at a particular phase of each step, a slightly different phase being characteristic of each cell. Their motor fields were all in the sole of the foot and depolarizing current evoked local longitudinal contraction, detectable when observed visually from beneath. Cells a and b evoked contraction about level with the siphon, a slightly more posterior than b. Cell c had the most anterior motor field, slightly anterior to the haemolymph-retaining clamp. The field of cell d lay under the anterior end of the visceral hump. Their phases of maximum discharge frequency are what would be expected during passage of a forwardly directed wave of longitudinal contraction (see Lissmann, 1945). (In Aplysia fasciata the wavelength of a step is long and the wave travels fastest in the middle of the foot, so that locomotion appears slightly leech-like with a single wave on the foot at a time. In Aplysia depilans this feature is considerably more marked and the foot is anatomically specialized to form an anterior and a posterior ‘foot’; these are alternately attached and detached.)
Not unexpectedly in a ‘hydrostatic’ animal, other neurones also discharge phasically during stepping. For instance, this is true of the parapodial levator motor cell in Text-fig. 8. Discharge frequencies are lower, however, and the parapodia of a creeping animal wave up and down only slightly with a period of 10–15 s. Presumably this activity is necessary to counteract the tendency of the parapodial sinuses to fill with haemolymph when the pressure in the pedal sinuses is increased by the contraction of foot muscles. A crawling animal normally holds its parapodia partially contracted and folded over the visceral hump. The effect on creeping of experimentally increasing the tension in the foot was not apparent, except perhaps slight excitation during stretching, but this could be a non-specific effect of deformation where the posterior hooks were inserted into the muscles of the tail. Text-fig. 9 (d) shows an apparent decrease in regularity of stepping when this one motor cell was hyperpolarized to prevent firing, but this effect was not consistently observed and would in any case hardly be expected as a direct effect since at a conservative estimate there must be at least 150 such longitudinal motor units in the whole foot (see Discussion). If there were recurrent axon branches from the motor cells, these might influence pre-motor interneurones with sufficiently large fields to produce such an effect; Dorsett (1968) did show the existence of numbers of neurones in the pedal ganglia of Aplysia punctata with ipsilateral peripheral axons (which could therefore have been motor cells) and having crossed axon branches ending in the contralateral pedal ganglion, so such an effect is at least possible.
Swimming: pedal neurones and the left pleural giant cell (LGC)
About 25% of preparations deprived of contact with the substratum exhibited recognizable swimming movements. Only in about half of these cases was the activity sustained for long enough to permit recordings of neuronal activity. In the best pre-parafions episodes of flapping could be evoked for a period of 6 h following the start if recording. The present observations are thus from a small number of preparations. Though caution is clearly necessary at this stage, they are included because among other things they do suggest one possible behavioural role for the left pleural giant cell (LGC). The almost uniquely extensive peripheral branching of the axons of this cell (at least nine major branches-Hughes, 1967) render its behavioural role of particular interest.
The swimming obtained from dissected preparations is shown in the flash photograph of Plate 1, fig. 2. The parapodia are turgid and fully extended, the gill is fully exposed and the foot long and narrow with the creeping surface folded in on itself. Text-fig. 10 shows the tonic activity of one of the motor cells with a transversely oriented motor field which fold the foot in this way during swimming. By contrast, during creeping the same animal was shorter and had a turgid foot with a wide creeping surface, the gill was covered and the parapodia were closed as far over the turgid visceral hump as the restraining hooks would allow (Plate 1, fig. 1). The only apparent differences in creeping and swimming between intact animals and dissected animals were that in the latter the periodicity of both kinds of locomotor movements was about two-thirds what would have been expected from an animal of this size, and that the tentacles were not extended forwards in front of the mouth in the characteristic posture during swimming, almost certainly because of the absence of a hydrostatic skeleton anterior to the haemolymph-retaining clamp.
Text-fig. 11 shows the activity of a large pedal cell with a contralateral peripheral axon (direct stimulation of cells of this type did not consistently evoke movements) and the LGC during (a) creeping and (b) swimming. During creeping both neurones discharged irregularly, but during swimming the pedal cell fired a burst corresponding to each beat, whereas the LGC discharged erratically with high-frequency bursts at intervals between episodes of regular swimming. Both cells were slightly hyperpolarized in (a) to show the changes in discharge frequency more clearly. It is of interest that, even when slightly hyperpolarized, the LGC in these preparations showed high-frequency synaptically evoked spike activity; in isolated nervous systems or under less good experimental conditions, it is a silent cell responding only phasically to depolarization. The beats of the parapodia were usually rather irregular for a while at the start of swimming, but this is exaggerated by distortion resulting from the large amplitude of the movements. The parapodia swing through nearly 1800 and even their anterior margins near to the visceral hump, where the transducers were attached, move several centimetres; when fully expanded they usually hit the Perspex strip supporting the neck clamp. The movements of the more posterior parts of the parapodia appeared to be the normal symmetrical metachronal beats seen in intact animals.
In an earlier communication (Hughes, Weevers & Hartley, 1969) certain rather slight effects of stimulating the LGC were reported-rather slow changes in longitudinal foot tension and accentuated periodic waving movements of the parapodia. At that time these effects were interpreted as heralding partial activation of a swimming response. However, a more injurious haemolymph-retaining clamp was then in use and it had not been realized that creeping movements (more readily evoked than swimming) resulted in slow periodic waving of the parapodia (see Text-figs. 8,9). The period of the movements then observed was much nearer to the normal period of a step than of a swimming beat. This new interpretation of the earlier results is reinforced by the experiment described below and illustrated in Text-fig. 12. It was noted that swimming tended to occur in a series of bursts interrupted by periods of longitudinal contraction of quite short duration. A series of these bursts of swimming lasting for up to 20 min could be evoked by allowing the animal to ‘rest’ with the foot in contact with the Polythene belt and then releasing this by passing a flat strip of Perspex between roller md foot so that the platform dropped. Sometimes the animal spontaneously released the platform, which fell under its own weight as swimming started. It was found that if the previously hyperpolarized LGC was depolarized, causing it to fire steadily at a fairly high frequency, such episodes of swimming were terminated within 10 – 20 s (Text-fig. 12). In one preparation this was repeated six times within a period of 2 h. Each time, after a variable latency, the sole of the foot became wide and flat and shortened markedly, and very frequently swinging movements of the foot would follow (see the slow levation of the right and depression of the left parapodium with accompanying longitudinal contraction of the foot shown in Text-fig. 12).
Thus the LGC may be one of the neurones involved in the change-over from the swimming to the creeping ‘mood’. Very likely other neurones are involved in the reverse change, since for about 10 min after stimulating the LGC in this way it is rather hard to obtain swimming, even from a freely suspended animal. Furthermore, not all preparations showed the inhibition of swimming so clearly, mostly because the spontaneous swimming episodes were shorter and it was uncertain whether these were terminated by discharging the LGC or spontaneously by some other means. The long variable latency and inconstant occurrence of this inhibitory effect indicates that it may be indirect or normally the result of several command neurones acting in concert. Furthermore there could be many reasons for stopping swimming, such as ‘hunger’, ‘tiredness’, ‘perceiving a mate’, etc.
DISCUSSION
The respective roles of motor cells and the peripheral plexus
Observations from the early literature on the role of the different gastropod ganglia (reviewed in Bullock & Horridge, 1965) make it clear that normal muscular function is impossible without the pedal ganglia, the pleural being less essential. In both Aplysia and Helix denervation of the pedal ganglia leads to progressive shortening of the sole of the foot, ending in death. Together with the observation that the shortening is delayed by reducing the amount of peripheral stimulation, this suggests an excitatory action of any peripheral motor cells which would normally be subject to inhibition by the pedal ganglia. Were it not for the rather enigmatic peripheral plexus, the present observations of single apparently motor cells in the pedal ganglion might be taken to indicate a direct route of excitation to muscle fibres of the kind described by Ramsay (1940) for Helix buccal retractor muscles. Willows (1968) described the same kind of directly evoked movements in Tritonia (unidentified cells near to cells 3, 4, 33, 34). He suggested that perhaps the peripheral plexus took care of relatively local and reflex responses. In the present work 15 % of sampled cells behaved like motor neurones. A conservative estimate of the number of cells in the pedal ganglion would be 2000 3000 on each side. Assuming no sampling errors this would therefore include 600-900 motor cells. A fully expanded 400 g Aplysia has a surface area of approximately 600 cm2, which would allow considerable overlap between motor fields 2 cm2 in area (about the average observed here) arranged in two overlapping sets oriented at right angles to each other. Such a complement would probably suffice for swimming movements, but would constitute a rather crude system, probably insufficient to co-ordinate the finer movements during creeping. Either the peripheral plexus must have a degree of autonomy in co-ordinating waves in the foot, or some of the cells found in the pedal ganglion having peripheral axons which yet evoke no apparent muscular contraction must be small field motor cells. The progressive shortening of the foot which results from denervation imposes a further degree of complexity on this picture and may imply peripheral inhibitory or ‘relaxing’ innervation, either of muscle or of other motor cells in the plexus.
The role of the pleural ganglia
Ten Cate (1928) believed the pleural ganglia of Aplysia to be without significance in the defensive movements of a parapodium which he evoked reflexly by contralateral stimulation. The present observation that discharging the LGC discourages swimming and promotes creeping suggests that this ganglion may be involved in activities of a more complex kind than these protective reflexes. There are also certain preliminary indications that other large pleural neurones may act as command cells.
Whether this is correct or not, the presence of peripheral axons, often widely distributed, from these cells poses a considerable problem. If there are pedal motor cells apparently serving as a final common path to at least some motor units, what function do the efferent pleural axons fulfil? Willows (1968) occasionally observed patterned bursting in small (motor cell?) somata in phase with normal swimming movements. He suggested therefore that these may be driven by the appropriate command cells; but at the same time he drew attention to the fact that the command cells themselves, as well as their follower populations, have axons that pass to the periphery in the nerve drunks appropriate to the various functions. It is hard to conceive that the information content of a series of bursts in one or a few command cells could be sufficient to coordinate directly a complex activity such as swimming. In any case, the pedal motor cells seem to be active at the same time, so any built-in co-ordinating circuitry of the pedal ganglion is likely to be activated in a manner subordinate to driving by appropriately connected command cells. Perhaps the peripheral axons of the command cells have an effect on the peripheral plexus. It may be no coincidence that both are features peculiar to gastropods.
ACKNOWLEDGEMENT
Valuable technical advice and many useful discussions with Professor G. M. Hughes are gratefully acknowledged. I am indebted to the University of Bordeaux for excellent laboratory facilities and accommodation at the Arcachon marine station, to the S.R.C. for financial support through a grant to Professor G. M. Hughes and to Bristol University for travelling expenses.