The Function of A Heteromorph Antennule in A Spiny Lobster, Panulirus Argus^*

Ablation of an eyestalk in decapod Crustacea is often followed by regeneration of a heteromorphic appendage in place of the amputated eye. The appendage usually copies an antennule; sometimes it includes basal segments and both flagella, at other times only the outer flagellum is present (Herbst, 1896). It is generally agreed that loss of the optic ganglia is essential for the heteromorphic regeneration, and that the new structure makes functional connexion with the central nervous system (Budden-brock, 1954; Herbst, 1900; Steele, 1907). The exact nature of such connexions remains obscure, however, and relatively few individuals have been studied. F rom the published accounts (Lissman & Wolsky, 1933; Herbst, 1910, 1916) there is some reason to think that variability may exist among and within species.

This paper describes the physiology of a heteromorph antennule found on a male spiny lobster, Panulirus argus Latreille, displayed at the Bermuda Aquarium during the summer of 1961. The heteromorph grew in place of the left eyestalk and was in the form of an antennular outer flagellum. The responses evoked upon stimulation of the heteromorph flagellum suggest that the patterns of central connexion of some of the fibres from the heteromorph were similar to those of fibres from the normal homolateral antennule. A brief summary has been published (Maynard, Cohen & Stephens, 1961).

The animal was an adult male P. argus Latreille. It weighed 1300 g. and was in good condition at the time of observation, 4–7 August 1961). After initial observations of behavioural reflexes, electromyograms were taken of muscular activity in the second and third segments of the normal antennules ; and, finally, intracellular recordings were obtained from motor neurons and interneurons within the exposed cerebral ganglia.

Normal behaviour and effects of chemical stimulation were observed with the animal either in a holding tank several feet long and approximately 30 in. wide, or in a smaller, glass-sided aquarium. Fresh, running sea water circulated through both tanks. Some reflexes were examined while the lobster was held in the hand out of water, or while tied to a board.

Electromyograms of muscular activity in the second and third antennular segments were made with the lobster tied to a board and partially immersed in sea water. Small cuts were made in the cuticle of the antennule and contact with the muscle was achieved by silver wire wound around the antennule at the point of the cut. Normal and heteromorph flagella were stimulated electrically though electrodes spring-clamped to the surface of the flagellum. Optic ganglia were stimulated through pins inserted on either side of the retina (Fig. 1).

Intracellular recordings from units in the cerebral ganglia were obtained with capillary microelectrodes filled with 3 M-KCI and led through a Bak cathode follower to a Tektronix 502 oscilloscope. For such recording the brain was exposed, de-sheathed, and perfused through the anterior aorta with oxygenated perfusion solution. The preparation remained alive for 12 hr. at about 23° C., and photographic recordings were obtained from ten penetrated units.

The heteromorph flagellum extended forward from a basal knob that apparently was derived from the original basal segment, basophthalmite, of the eyestalk (Fig. 2). The flagellum contained seventy-seven annuli and was 4·7 cm. in total length. Of the terminal forty-eight annuli (2 cm.) all but the last nine were equipped with guard hairs, companion hairs, and aesthetasc hairs characteristic of the outer flagellum of the normal antennule (Laverack, 1964; see also, Herbst, 1900). Both annuli and sensory hairs were distributed more irregularly than in a normal flagellum, but in general the external anatomy of the heteromorph flagellum was qualitatively identical with that of the outer flagellum of the normal antennule.

Afferent fibres from the receptors of the heteromorph apparently enter the brain along the path of the former optic tract on the eyeless side. The anatomy of the terminations of such fibres within the brain was not determined. The musculature at the base of the heteromorph flagellum was abnormal, but there was no evidence of functional muscles specifically associated with the heteromorph. The flagellum was incapable of independent movement, but it could be raised or lowered in concert with the right eye because of its attachment to the common interocular yoke (Fig. 3, Pl. 1). This contrasts with the condition in crayfish (Hofer, 1894; Lissmann & Wolsky, 1933), where nearly complete heteromorph antennules with basal segments and both flagella may regenerate and where apparently independent movement of the heteromorph has been observed.

The external structure of the heteromorph antennule is consistent with its being a mechano-or chemoreceptor, but the flagellum is so placed that its afferent fibres approach the brain in the region of normal optic tracts. Consequently it is difficult to anticipate the exact form of the response to stimulation of the heteromorph flagellum, or to be certain, a priori, that it is functional at all.

It was immediately apparent, however, that the flagellum was sensitive to stimulation. When gently tapped on the distal half of the heteromorph flagellum with a glass rod, the unrestrained lobster moved back and down to the right, withdrawing from the stimulus. The response was usually rather slow, and often appeared only after several taps. Movement of the rod immediately in front of the heteromorph was ineffective, so mechanical distortion caused by touch was presumably the effective stimulus. Similar, although more vigorous, withdrawal responses were elicited by mechanical stimulation of the right eye or eyestalk, or by repeated mechanical stimulation of either antennule (see Maynard & Dingle, 1963). The direction of withdrawal in each case indicated localization of the stimulus to the appropriate quadrant of the body. Stronger taps on the heteromorph flagellum of the quiescent lobster sometimes evoked generalized startle responses involving both antennae and walking legs. These were similar to startle responses evoked by taps on the eyestalk.

Attempts to demonstrate chemoreceptors in the heteromorph gave ambiguous results. In the normal animal chemical stimulation of an antennular flagellum usually leads to feeding activity. This response is lost, however, when the eyestalk on the side of the stimulated antennule has been amputated (Maynard & Dingle, 1963). Consequently the fact that no clear feeding responses occurred when fish juice was pipetted over the heteromorph flagellum does not necessarily indicate absence of peripheral chemoreceptors in the flagellum. When fish juice was squirted with sufficient force to cause passive movement of the flagellum tip, the withdrawal response appeared and was more vigorous than that occurring when sea water alone was squirted. This would suggest that input from chemoreceptors in the heteromorph was active in potentiating the withdrawal. More recently, in other lobsters, Laverack (1964) has demonstrated the presence of chemoreceptors in heteromorph flagella by electrophysiological means.

These observations establish the presence of functional connexions between the heteromorph and the central nervous system. The responses described are general ones, however, and can be evoked by a number of cutaneous stimuli. Consequently they do not tell whether input from the heteromorph antennule is treated centrally as though from retinal, from eyestalk, or from antennular receptors. To prove the latter, stable and specific reflexes are necessary ; and in order to explore central mechanisms these should be demonstrable in a restrained animal. Three reflexes proved suitable ; these were observed in a normal control as well as in the experimental lobster.

The adequate stimulus for the depression reflex in the restrained, normal lobster was gentle manipulation of the terminal portion of the outer flagellum. The response itself was a smooth depression of the stimulated outer flagellum and the next proximal (third) antennular segment.

Flagellar depression was particularly consistent unless there were special circumstances, such as general appendage rigidity of the hand-held animal. In normal lobsters this reflex appears to be specific for touch of the outer flagellum and does not occur for other stimuli. Similar flagellar depression to tactile stimulation has been reported in crayfish (Lissmann & Wolsky, 1933). In some cases it may represent initial, aborted movements which, if continued, would lead to cleansing of the flagella with the third maxillipeds.

The adequate stimulus for this reflex in the normal restrained lobster was manipulation of the inner flagellum. The response was a smooth elevation of the outer flagellum of the stimulated antennule. The inner flagellum is not supplied with a muscle, and cannot be moved independently.

Upon manipulation, touch, or tap of the eyestalk, brief bilateral antennular movements followed by bilateral elevation of the outer antennular flagella occurred. This response was rather variable compared to the flagellar elevation or depression reflexes, and showed rapid de-facilitation or habituation. In some cases this reflex was associated with more generalized activity involving all appendages.

It is highly significant that manipulation of the heteromorph flagellum also elicited ipsilateral depression of the normal outer flagellum. In fact, the response to heteromorph stimulation was more intense than that to normal flagellum stimulation. The heteromorph antennule, as indicated above, was incapable of independent movements itself. In no case did stimulation of the heteromorph flagellum elicit the eyestalk-flagellar or flagellar elevation reflexes, nor did eyestalk or visual stimulation ever cause flagellar depression as described above. One may conclude at this point that, in so far as the reflexes we have examined are concerned, the lobster does not ‘see’ with the heteromorph flagellum, and that the regenerated structure copies the antennular flagellum in function as well as anatomy.

Similar movements, however, need not always be caused by identical patterns of muscle activity. This is particularly so in decapod Crustacea, where antagonistic muscles are usually multiply innervated, and where peripheral inhibition at the muscle itself may occur. Even similar outer flagellar movements, which are mediated by a single, multiply innervated muscle acting against an elastic cushion at the joint, could conceivably be caused by several patterns of discharge in inhibitor and excitor motor fibres. Accordingly, we placed recording electrodes on the depressor muscle of the third segment (m. reductor₃), and then on the single muscle depressing the outer flagellum (m. reductor₄) in each antennule and re-examined the reflexes (Balss, 1941).

Positions of various recording and stimulating electrodes are shown diagrammatically in Fig. i. Records were taken with electrodes 2r on 4 August. Most potentials recorded were similar in amplitude and form and presumably originated in m. re-ductor₃. There were also some of smaller amplitude and a different but unknown origin (see ‘early burst’ in Fig. 9, Pl. 2). The lobster was then returned to the holding tank, and records from m. reductor₄ (via electrodes 3r) were made on 5 August.

Fig. 4 shows typical examples of the reflex discharge in the m. reductor₃ of right and left antennules during manual manipulation of flagella and eyestalk. These responses are summarized in Table 1.

In general these recordings confirm the behavioural observations. The response to stimulation of the heteromorph resembles more the response to stimulation of the outer flagellum than the responses to manipulation of either the inner flagellum or the eyestalk. The responses are not identical, however, for the heteromorph was (1) less effective contralaterally, and (2) often elicited greater absolute ipsilateral discharge rates than did the outer flagellum on the left side.

Fig. 5 shows typical reflex discharges in the m. reductor₄ following manipulation of the outer flagella, heteromorph flagellum and eyestalk. Only activity in the spsilateral muscle increases following flagellar stimulation; there appears to be no maintained contralateral effect. The response to eyestalk stimulation is brief in duration and bilateral, and shows no resemblance to responses to flagellar stimulation. The effects of inner flagellar stimulation are not shown but, as with m. reductor₃ (Fig. 4), caused ipsilateral depression of muscle-potential frequency.

The responses of m. reductor,, to single electrical stimuli applied to the flagella are shown in Fig. 6. Again there appears to be no contralateral response. The ipsilateral response is complex. An initial brief burst of muscle potentials is followed by a silent period that begins 90–100 msec, after the stimulus and lasts 80–120 msec. ; this in turn is followed by a second high-frequency burst that grades into a prolonged, high-frequency after-discharge. The latency of the first burst is approximately 50 msec., suggesting afferent conduction velocities of at least 7 m./sec. The response to stimulation of the heteromorph flagellum is simpler. Both the first and second burst appear lacking, and after a latency or silent period of 80–100 msec., a prolonged high-frequency response occurs. These temporal differences between the response to normal and heteromorph flagellar stimulation are due in part to differences in afferent conduction distances (Fig. 1). Other differences such as the initial burst in response to normal flagellar stimulation and the excessive late response to heteromorph input appear to have other causes, but the reflex activity evoked by the heteromorph continues to resemble that caused by normal outer flagellar stimulation more closely than any other examined here.

The response of m. reductor₄ to electrical stimulation of the different fibres was similar to that of m. reductor₄ (see Fig. 7).

On 7 August the lobster was opened, the brain was perfused and desheathed, and the antennules were amputed at the basal joint. The stumps of the antennular nerve were split into two bundles : the medial bundle contained large afferent fibres from the statocyst region at the base of the antennule; the lateral or flagellar bundle contained larger mechanoreceptor afferents from proprioceptor organs at the antennular joints (Wyse & Maynard, 1963), motor and inhibitor efferents, and numerous other fibres from flagellar hairs and sense organs. There were, therefore, five cerebral inputs available for electrical stimulation; ipsi-and contralateral medial and flagellar bundles and the heteromorph flagellum (Fig. 8). Only the input from the heteromorph was the same as that used in the preceding sections. The flagellar bundles contained fibres from the outer flagellum, but also included, as described above, a heterogeneous population of fibres from other antennular receptors and the inner flagellum.

Twenty units were penetrated. Ten were inactive and were unresponsive to all inputs, even though they had resting potentials ranging from 60 to 95 mV. The remaining ten responded to at least one and usually several of the available inputs, or they showed spontaneous activity, or both (Table 2). Four of these ten, of which two were considered to be motor neurons (Units 2 and 3) and two to be interneurons (Units 6 and 10), responded to heteromorph flagellar stimulation.

Fig. 9, Pl. 2, shows the response to stimulation of the heteromorph flagellum (Fig. 9 A, Pl. 2) and of the outer flagellum or the flagellar nerve bundle (Fig. 9B, Pl. 2) as recorded in the ipsilateral antennular nerve (traces 1 and 3), in two intracerebral motor neurons, (trace 2 represents Unit 2 of Table 2; trace 4 represents Unit 3 of Table 2) and in ipsi-and contralateral m. reductor₃ (traces 5 and 6).

The first and third traces in Fig. 9 A, Pl. 2, represent activity in efferent neurons recorded with extracellular electrodes placed on the ipsilateral antennular nerve about 8 mm. from the brain (Fig. 8). There was a spontaneous output. Following a single stimulus to the heteromorph flagellum a burst of efferent spikes occurred after a latency of about 35 msec. The burst lasted 32–34 msec, and was followed by a silent period of about equivalent duration before a late discharge period began about 100 msec, after the initial stimulus. A small burst of impulses sometimes occurred at the beginning of the late discharge, but this was not always obvious. The raised frequency of the late discharge lasted for well over 500 msec.

The second and fourth traces in Fig. 9 A (Pl. 2) represent intracellular recordings from each of the two motor elements. Traces 1 and 2, and 3 and 4 were taken simultaneously. Both units had resting potentials in the neighbourhood of 45 mV., both showed small spontaneous ‘spikes’ and synaptic noise, and both responded to heteromorph flagellar stimulation after a latency of 26 msec. The two motor units responded to the stimulus with somewhat different discharge patterns, however. Unit 3 (trace 4) will be considered first because its spike output definitely contributed to that recorded extracellularly in the ipsilateral antennular nerve (trace 3). An extracellular impulse (marked ‘. ‘) follows each intracellular spike with a latency of about 2·5 msec. ; the conduction velocity in this particular axon must therefore be about 4·5 m./sec. Unit 3 first responded to the stimulus with a prolonged, complex hyperpolarization or IPSP. A maintained depolarization which elicited a higher-frequency repetitive spike discharge then appeared after about 145 msec. Unit 3 therefore contributed spikes to the late discharge, but not to the initial burst, recorded in the antennular nerve.

The response of Unit 2 was more complex. Its spikes were not synchronized with any recorded in the antennular nerve, so its peripheral axon was probably damaged or destroyed during preparation, or else left the antennular nerve at a point proximal to the recording electrodes. Its activity, however, clearly synchronized with both the initial and later burst and with the late maintained discharge. In contrast to Unit 3, the initial response was depolarizing and excitatory, an EPSP. Although there appears to be no net hyperpolarization, several abrupt drops in potential suggest superimposed inhibitory effects, and it is likely that the return toward resting potential between the bursts represents inhibitor activity.

The fifth and sixth traces represent activity in the ipsi-and contralateral m. re-ductor₃ respectively. These records were taken 3 days earlier during the initial electromyography, but are included here to permit some correlation between the behavioural reflexes initially described and intracellular recordings. There is striking similarity with respect to the onset of inhibition and the late discharge between the ipsilateral muscle activity and Unit 3 activity. Equally apparent, however, is the absence of muscle activity correlating with the initial impulse burst in the antennular nerve. Presumably fibres involved in the initial impulse burst innervated proximal muscles of the antennule which were not examined electromyographically.

Fig. 9 B (Pl. 2) presents the same sequence of records as Fig. 9A (Pl. 2) but shows the response to stimulation of the ipsilateral flagellar bundle in the first four traces, and of the ipsilateral outer flagellum in the last two traces. The efferent discharge patterns recorded in the flagellar bundles (traces 1 and 3) differ quantitatively from those in Fig. 9 A (Pl. 2). The latency of the first burst was about 7 msec, rather than 35 msec., and the late maintained discharge was much less intense and briefer than that obtained following heteromorph stimulation. The initial burst itself also appeared double, with two periods of’maximum’ spike frequency. In other ways, however, the total pattern was very similar. There was an early and a late burst, and the late burst was followed by a maintained, though small, increase in discharge frequency. The interval between the two bursts, 60–67 msec., was about that observed in the response to heteromorph stimulation. With repeated stimulation, however, the late burst tended to fatigue and disappear.

In Unit 3 the initial response to flagellar bundle stimulation was a depolarizing post-synaptic potential with an antidromic ‘spike’ that had a latency of about 2 msec. About 7 msec, after the stimulus, the first hyperpolarizing response began. This presumably a compound IPSP, immediately cut off the depolarizing EPSP and was followed by a prolonged, complex hyperpolarization. There was some indication of an EPSP at the time of the second burst, but maximum depolarization and spike discharges occurred only at the time of the late maintained discharge. In Unit 2 the minimum latency according to records with higher temporal resolution was 2·8 msec. The initial response, though depolarizing as in Fig. 9 A (Pl. 2), was quite different in form and time course. It rose abruptly to about 40 mV., presumably indicating synchronized EPSP’s. About 7 msec, after the stimulus the EPSP began an abrupt repolarization which eventually passed slightly beyond the initial resting potential, presumably the result of a synchronous inhibitory input. Following this a complex pattern of potential changes occurred that was very similar qualitatively to that following the initial response to heteromorph stimulation (Fig. 9 A, Pl. 2). In both Unit 2 and Unit 3, therefore, there was a brief preliminary excitation-inhibition sequence that began before and appeared to overlay slightly the initial part of a more prolonged late response. This latter response, in both neurons, was essentially identical with the entire response elicited by heteromorph activation. The primary difference in the responses to normal and heteromorph flagellar stimulation thus appears to be the addition of an early component in the response to stimulation of the normal flagellum.

The muscle responses recorded on the last two traces cannot be directly compared with the nerve activity above because the outer flagellum rather than the flagellar bundle vas stimulated. The input path of these recordings consequently was about ten times longer, with a corresponding increase in response latency, and the input was more homogeneous because only fibres going to the outer flagellum were stimulated. Nevertheless, the initial inhibition and the late discharge are apparent. An initial burst in the form of two small muscle potentials is also present. It must be emphasized that, although this initial muscle potential burst may correlate with the first portion of the initial efferent burst recorded in the antennular nerve, it cannot be equated with the entire double burst. Following heteromorph stimulation the later portion of the initial nerve impulse burst (traces 1 and 3) occurs without concomitant recorded muscle activity.

Table 3 gives estimated latencies and distances for the preparation and activity as recorded above. These data permit calculation of ‘gross mean conduction velocities ‘that include central synaptic delays. Although crude, such values suggest that the nerve bundles from the antennular outer flagellum contained a population of fibres whose conduction velocity was either greater or whose connexion with motor elements was more direct than any fibres found in the nerve from the heteromorph flagellum in this preparation. The early portions of both the initial burst of nerve impulses and the initial muscle potential burst following normal flagellar stimulation require such fast afferents, suggesting that some of the major differences between the effects of heteromorph and normal flagellar stimulation could result simply from the absence of this particular fast, direct fibre bundle in the heteromorph.

Figs. 10 (Pl. 3) and 11 (Pl. 4) show the responses of a presumed interneuron (Unit 6) that responded to all five antennular and flagellar inputs and also gave brief responses to sudden changes in illumination. The upper trace in each record represents efferent activity recorded extracellularly in the ipsilateral antennular nerve. The lower trace records simultaneous intracellular activity. The resting potential of the interneuron was about 53 mV.; it developed relatively large, 15–20mV. depolarizing post-synaptic potentials but only small spikes, 4–5 mV. The ipsilateral flagellar bundle was the most effective input, eliciting both an early, relatively brief, complex EPSP and a later, more prolonged, depolarizing complex; both early and late EPSP’s were large enough to trigger spikes. With repetitive stimulation at 5–10 per sec., the early response showed de-facilitation, while the late response appeared to facilitate within a few stimuli, producing a maintained depolarization and a repetitive spike discharge that reached instantaneous frequencies of over 200/sec. With continued stimulation the facilitated response gradually declined as shown in Fig. 11 (Pl. 4).

Of the other three antennular bundle inputs, that of the contralateral flagellar bundle was most effective and was the only one to initiate spike potentials. The ipsilateral medial bundle was next, and the contralateral medial bundle was the least effective. In all instances an early, de-facilitating response occurred. With contralateral flagellar input there was also a relatively small, apparently facilitating, late discharge. The intensity of efferent activity in the ipsilateral antennular nerve paralleled intracellular activity. In summary, ipsilateral input was more effective than an equivalent contralateral input, and on either side flagellar bundle input produced a larger, more complex response than medial bundle input.

The response of this interneuron to a single stimulus to the heteromorph flagellum was slight: a small EPSP after a latency of 30–40 msec., and possibly, later EPSP’s. With repetitive stimuli facilitation occurred, and after 3–5 stimuli at 5–10 per second a large maintained depolarization with evoked spike potentials developed. As with the facilitated response to ipsilateral flagellar bundle stimulation this declined with continued stimulation. The efferent fibre discharge also paralleled the intracellular activity. Although the response to heteromorph flagellum stimulation clearly differs from all antennular bundle inputs in the absence of an early response, it is very similar to the late component of the response initiated by the ipsilateral flagellar bundle.

The responses of another presumed interneuron (Unit 10) are shown in Fig. 12, (Pl. 1). The unit had a resting potential of about 60 mV. and developed spike and post-synaptic potentials of about 45 and 25 mV. respectively. Early EPSP responses to each of the four antennular bundle inputs were present. The amplitude of the recorded EPSP’s initiated via ipsilateral input was characteristically greater than those initiated by contralateral input, but the critical voltage for spike initiation was greater with ipsilateral than with contralateral input. Consequently this unit contrasts with Unit 6 (Fig. 11, Pl. 4) in that contralateral input was the most effective in eliciting spike discharges. In some cases de-facilitation also seemed more rapid with ipsilateral input. With large input volleys late responses in the form of scattered EPSP were present. These were very small when compared with the early, more synchronous, compound EPSP.

Stimulation of the heteromorph flagellum did not elicit an early response in this interneuron; in fact a single stimulus was apparently completely ineffective. With two or three repeated stimuli, however, scattered EPSP occurred, and with continued stimulation these occasionally avalanched to reach spike threshold as shown in Fig. 12, (Pl. 1). The appearance of such EPSP roughly corresponded with the appearance of large efferent activity in the ipsilateral antennular nerve. As with Unit 6 (Fig. 11, Pl. 4), the early maximum response was not maintained, but with still further stimulation declined to a continued, asynchronous barrage of EPSP.

In all four penetrated neurons, therefore, intracellular recordings of responses to heteromorph stimulation are very similar to certain components of responses to flagellar bundle input. The similarities between such flagellar and heteromorph responses in any one unit appear greater than similarities between heteromorph responses of any two of the four units. Such correspondence makes it necessary to speak of characteristic flagellar responses that vary in detail according to the nature of the post-synaptic unit rather than to speak of characteristic heteromorph responses which are independent of the nature of the interneuron and its normal connexions.

It is important to know whether the penetrated units responding to heteromorph input were located in their usual positions within the lobster brain, or whether they were all neighbouring elements in one locale.

Electrode penetrations were directed into the antero-dorsal face of the cerebral ganglia. Electrode tip depths beneath the surface were measured for each recorded unit by means of a micrometer scale on the electrode drive. At the end of each penetration, the electrode was cut and left in place in the brain. After the experiment the brain was fixed in acetic-picric acid in sea water. The sites of the electrode penetrations on the brain surface were sketched and the electrodes were removed (see Fig. 13 C). The brain was then embedded in paraffin and finally sectioned at right angles to the electrode track. The tracks were visible as holes in the section, and were followed in serial sections through the brain. With a correction for shrinkage and distortion during preparation, the approximate area of the brain in which any given recording was made could be estimated.

Figs. 13 A and 13 B show sample tracings from sections at the approximate level of recording for several of the ten penetrated units, and Fig. 13 D, a sagittal reconstruction in the approximate plane of the penetrations. Positions of the four units giving responses to heteromorphic stimulation are indicated. The two motor units were located in the dorsal portion of the antennular neuropil (parolfactory lobe ; Hanstrom, 1947). The first interneuron was found in a generalized neuropil area dorsal and lateral to the motor elements. The second interneuron occurred more dorsally and nearer the mid-line very close to a number of commissural fibres. Perhaps this explains the unusual effectiveness of contralateral input for this second interneuron. There is no evidence of unusual aggregation of the units responding to heteromorph input. In so far as can be seen, they were located in normal positions and were interspersed with neighbouring elements which did not appear to respond to heteromorph input.

Undoubtedly the afferent fibres from the heteromorph flagellum of the lobster described here made functional connexions within the cerebral ganglia. Both Herbst (1910) and Lissmann & Wolsky (1933) also found functional heteromorphs in decapod Crustacea, and both reported general responses to heteromorph stimulation. Lissmann & Wolsky observed additional behavioural responses with components similar to those initiated by normal antennular stimulation; Herbst did not. Our present concern, therefore, is with the nature of heteromorph connexions. In this individual Panulirus behavioural observations were supplemented with direct electrical recording at various points of the reflex pathway. Consequently we are able to bring more direct and more detailed evidence to the question : to what extent are the patterns of intracerebral connexions of afferent nerve fibres from the heteromorph flagellum identical with those of afferent fibres from the normal, ipsilateral antennular flagellum?

It is useful to distinguish between two kinds of responses to normal flagellar stimulation, general and specific. The general response, as illustrated by the withdrawal and startle responses, involves an appreciable portion of the efferent neural apparatus and may be elicited by input from any of a number of receptors over the body surface. The response to various inputs differs primarily with respect to direction of movement, but the ‘place sense* or local sign may be poor so that behavioural differences need not be obvious so long as the input comes from the same general quadrant of the animal. In contrast, the specific response, illustrated by flagellar depression, involves not only a small number of efferent neurons and very specific movements, but also an effective stimulus normally limited to a discrete population of receptors located in one specific region.

An argument for the identity of two inputs based on their ability to evoke similar specific responses is more compelling than one based on ability to evoke similar general responses. Therefore, although apparently similar withdrawal responses are initiated by heteromorph and normal ipsilateral flagellar stimulation, for the present more emphasis will be placed on the specific flagellar depression response. At the very least, the observation that stimulation of the heteromorph evokes movement like that of the ipsilateral depression response rather than elevation, or some other specific activity, implies that some of the central connexions from the heteromorph are more similar to those of the ipsilateral normal antennular outer flagellum than to any of those from the eyestalk, the ipsilateral antennular inner flagellum, or from either of the contralateral flagella. The following paragraphs will consider the extent of this similarity.

When stimulation of normal and heteromorph flagella were strictly comparable, either in the form of manipulation or electrical stimulus, the evoked flagellar depressions were very similar but not identical. There were two major discrepancies: first, the duration and extent of flagellar depression or of the late discharges in the depressor muscles tended to be greater following heteromorph than normal flagellar stimulation; and, secondly, following electrical stimulation of the normal flagellum there was a brief, early burst of muscle potentials that was absent following heteromorph stimulation. The discrepancies remained when stimulation of the median bundle of the antennular nerve, a more heterogeneous input, was substituted for the outer flagellum, and were observed in the responses of both motor neurons and interneurons.

If these discrepancies are ignored, however, the similarities between the responses to the inputs are very impressive. They include : the pattern of movement of the outer flagellum and third segment of the antennule; the early and late excitation found in both efferent motor fibres and in a motor neuron ; the late IPSP-EPSP sequence in another motor neuron; and the late facilitating excitation observed in two different kinds of interneurons. In addition, the efferent activity elicited by both inputs was apparently limited to the same few motor neurons : there were no irrelevant movements.

These similarities seem sufficient to suggest that convergence of the relevant afferents from heteromorph and normal flagellum occurs at some stage prior to the final efferent neuron. Presumably both inputs converge on the postulated internuncial pool whose activity accounts for late inhibition—excitation PSP sequences that far outlast the arrival times of impulses in the slowest of the afferent fibres.

This hypothesis is presented as a diagram in Fig. 14. To the left is a flow chart showing functional stages ; to the right a diagram of a minimal anatomical substrate. There is direct evidence for the stages leading back from flagellar movement to the input stage of the motor neuron (or interneuron). The remaining stages between the motor neuron and the stimulus are more obscure.

The pattern of late PSP recorded in motor and interneurons indicates that the output from the postulated intemuncial pool must involve spatially co-ordinated, sequential activity in several inhibitor and excitor neuron populations. Since the exact balance of inhibitor–excitor input activity differed for all four neurons examined, and yet was similar for both normal and heteromorph inputs in any one neuron, a rather complex and specific output is implied. The origin of the pattern is uncertain; it could result from a series of relay elements with different temporal properties as shown in Fig. 14, but more probably it also involves various feed-back networks. In any case at least two, and more probably three, populations of interneurons are required. In Fig. 14, the population E₂ may be considered responsible for the later part of the first excitatory bursts ; I₂ responsible for the con-current or subsequent inhibition; and E₃ for the late, long-lasting excitation. Neither of the penetrated interneurons can be assigned to any of the above populations, and like the motor neurons they must represent later stages in the response.

It should be re-emphasized, perhaps, that the diagram of the intemuncial pool is primarily intended to demonstrate that even with a simplified anatomical substrate the input pattern to the motor neuron is more complex than the output and requires multiple interneuron types. Undoubtedly the actual patterns of connexion in the lobster are much more complicated and probably include direct monosynaptic afferent-efferent junctions and recurrent influences from the motor neurons themselves. The few experiments involving interaction between heteromorph and normal flagellar input that might have given more information about the nature of convergence in the interneuron pool were inconclusive.

One of the discrepancies mentioned above, that of the unequal strength of the depression reflex following normal and heteromorph stimulation, can be explained by assuming excessive activity of E₃ interneurons following heteromorph stimulation. The exact mechanisms involved seem less certain. Possibly the heteromorph input is stronger because of more extensive or larger presynaptic ramifications ; or perhaps the excitability of E₃ neurons is greater because of reduced inhibition. Certainly, however, this inequality of the depression response need not imply major qualitative dissimilarities in the central pathways, nor require that additional or different interneuron populations be involved.

The early response of excitation-inhibition found with normal input but lacking with heteromorph input is difficult to account for within the confines of the internuncial pool described above. It is reasonable to assume that the normal input includes a bundle of fibres that makes other functional central connexions. The heteromorph input presumably either completely lacks such afferent fibres, or if present they fail in this lobster to make proper, effective central connexions. The latter explanation is perhaps preferable because there are no obvious missing or rare receptors on the heteromorph (Laverack, 1964).

The short latency of the early response, 2·8 msec., suggests an afferent conduction velocity of at least 9 m./sec. (Table 3) and leaves very little time for transmission across a relay intemuncial. Accordingly excitation in the ‘early pattern’ is shown as monosynaptic, while inhibitory interneurons, activated from collaterals of the afferent fibre are proposed as an obvious mechanism to account for the early, synchronous inhibitory effects. Of course, other schemes such as recurrent inhibition or direct monosynaptic inhibition by slower afferents are also possible.

No direct recordings were made of the input volley in this animal, but there is relevant information from other preparations. In normal outer flagella, single maximal electrical stimuli similar to those used here generally elicit single volleys without repetitive discharge. The conduction velocities of the afferent fibres fall into two major groups, a faster component with velocities ranging from about 8·5 m./sec. down to 3 m./sec., and a much slower component with average conduction velocities of about 0·5 m./sec. Such a complex input volley would be expected from the normal flagellum in the present preparation. Since the peripheral receptor structures on the heteromorph do not differ materially from those in the normal flagellum, comparable similarity of the afferent fibre population arising from the receptors might also be expected. Latency measurements show, however, that conduction time via heteromorph input to the motor neuron is more than twice that for a comparable distance via normal antennular input. This increased time possibly results from delay at an interposed group of relay interneurons (indicated by ‘in Fig. 14), but it seems more probable that the newly generated afferent fibres from the heteromorph have conduction velocities much less than those of normal flagellar afferents.

Our observations do not give sufficient information about the relative importance of the various afferent components in eliciting the responses described above. Apparently, however, characteristic responses begin before the arrival of impulses in the slowest component, suggesting that the latter may be of lesser importance than more rapidly conducting fibres. Since the slow fibres are considered to arise in neurons supplying the aesthetasc hairs rather than mechanoreceptors, the responses considered here cannot be proper olfactory or chemosensory reflexes.

A simple answer to the question of whether the central connexions of afferents from heteromorph and normal flagella are identical does not seem possible. The input from both structures is heterogeneous and the central connexions are correspondingly diverse. At least four afferent components must be considered. First, there is no evidence in this animal about the nature of the central connexions of the slowly conducting fibre component presumably originating in the aesthetasc hairs of the flagella. We can say nothing, therefore, about one of the major afferent components of the flagellar nerve, or even affirm that it exists as a functional pathway in the heteromorph nerve. Secondly, a rapidly conducting afferent component present in the normal flagellar nerve is absent or non-functional in heteromorph input. For this fibre group, at least, the central connexions of the two inputs clearly differ. Thirdly, one afferent fibre component, possibly including that derived from non-specific tactile receptors, terminates centrally in generalized defence and withdrawal centres. Behaviour evoked by it is appropriate to the stimulus and to the stimulus site. Qualitatively, movements originated via this particular heteromorph input do not differ appreciably from those initiated by eyestalk stimulation in a normal animal. Although not observed in this animal, in other specimens of the same species (Maynard, unpublished observations ; see also, Herbst, 1910; and Lissmann & Wolsky, 1933) directed cleaning movements show discrimination between the site of a stimulus to the heteromorph and that to the normal flagellum. There is no reason to believe that the particular patterns of connexion made by this afferent fibre component have any more similarity to flagellar input patterns than those from any other region of the anterior ipsilateral quadrant of the animal. Fourthly, another fibre component, and the last to be considered here, terminates in regions of the brain responsible for specific, oriented responses of the normal antennule. Such responses, however, are completely inappropriate in so far as the geographical position of the heteromorph receptor elements is concerned. They involve movements of the normal ipsilateral antennular flagellum appropriate for flagellar, not heteromorph, stimulation. As discussed in detail, the similarity of the response evoked by heteromorph and normal flagellar stimulation is not superficial, but involves nearly identical activity in a variety of central elements so that the inference of similarity of central morphological connexions seems well founded. Assuming identity of receptor types in the heteromorph and normal outer flagellum, these connexions appear appropriate for the receptor type, not the receptor location. Only with respect to fibres eliciting the flagellar depression response, therefore, can the answer to the question of the identity of central connexions be a reasonably firm yes. With respect to all other components, we must say either ‘we have no evidence ‘, or ‘connexions are not identical’.

Although similar, our observations as summarized in the above paragraphs differ in detail from those of both Herbst (1910) and Lissmann & Wolsky (1933). Herbst examined shrimp (Palaemon serratus and P. rectirostris) and spiny lobster (Palinuros vulgaris). In both genera chemical and mechanical stimulation of the heteromorphy elicited withdrawal and local cleaning of the stimulated area by the appropriate walking legs. This response indicates an accurate local sign and was similar to that caused by eyestalk stimulation, but completely different from the activity following stimulation of a normal antennular flagellum. When stimulated mechanically or with strong or irritating chemicals the antennular flagella were lowered, grasped by the third maxillipeds and cleaned. Herbst’s observations therefore show general responses, but no specific antennular reflexes, following heteromorph stimulation. It should be noted, perhaps, that only one individual of Palinurus was examined, and it was tested only 4 days after moult when excessive sensitivity to irritants might be expected (see Maynard & Dingle, 1963). Lissman & Wolsky (1933) also found local cleaning with walking legs at the site of stimulation in the single crayfish (Potamobius leptodactylus Eschh.) they studied. In addition, however, mechanical and particularly chemical stimulation of the heteromorph evoked cleaning responses of the normal antennule with third maxillipeds. Such specific, misdirected cleaning represents a more complete and complex response than the depression described for Panulirus, and in this particular crayfish involved the contralateral rather than the ipsilateral antennule. The variability often found in regenerated systems, however (see Gaze & Jacobson (1963) for recent examples) and the limited and essentially anecdotal nature of the evidence make evaluation of such apparent interspecific differences difficult. Rather we would suggest that these observations help confirm our impression of at least two receptor classes in the heteromorph, one forming central connexions on the basis of peripheral ‘local sign’, the other ignoring local sign but forming connexions on the basis of receptor modality or type.

We do not know the history of the lobster described here, so the age of the heteromorph and the consequent time available for neural reorganization is uncertain. Possibly the heteromorph dates from larval or early post-larval stages, but in form it resembles heteromorphs produced experimentally in much briefer times, and it seems very likely that the eyestalk was lost at an adult stage. Of a number of experimental lobsters of similar body size, two developed regenerate heteromorphs of 4·0 and 4·1 cm. respectively within two moults and 321 days after eyestalk removal; one a heteromorph of 4·1 cm. within four moults and 687 days post-operation time ; and one a heteromorph of 4·9 cm. within three moults and 526 days post-operation time. If the rates of regeneration of these animals maintained in laboratory tanks were similar to that of the animal described here, then the heteromorph length of 4·7 cm. suggests that it is at least between two to four moults and 1–2 years old. This is an important point because it implies that the ingrowing afferents of the regenerate heteromorph entered an adult, organized, functioning neuropil.

The mechanisms involved in the establishment of such specific, apparently maladaptive patterns of connexion in the nervous system have received attention for a number of years. The problems were perhaps first clearly enunciated by Weiss in 1936, and again reviewed by Sperry in 1951. A number of investigations employing supernumerary appendages or abnormal regenerates, principally in amphibia and fish (see Sperry, 1951), but also in birds (Székely & Szentagothai, 1962), give evidence for the existence of neuronal specificity in vertebrates and in general lead to the conclusion that mutual chemical compatibility or recognition between specific neuron types must exist. Except for some of the recent work of Gaze and co-workers (Gaze & Jacobson, 1963 ; Gaze, Jacobson & Székely, 1963), Maturana, Lettvin, McCulloch & Pitts (1959), and Eccles (summary in 1964), however, most such conclusions regarding neural specificity are inferred from behavioural observations. The observations reported in this paper do not materially further our understanding of the developmental processes involved, but they do demonstrate, with a directness not usually available, the preciseness and specificity of the inter-neuronal activity patterns that may result from such processes. In addition, by confirming the essential aspects of the earlier report of Lissmann & Wolsky (1933), our observations help to provide examples of the concept of neuron specificity in a non-vertebrate phylum, the Arthropoda.

When comparing these results with those of others, however, several points should be remembered. First, the afferent fibres of the heteromorph presumably arise in new sensory cells formed during regeneration and grow into the brain along paths that never, under usual circumstances, carry fibres from that particular kind (or kinds) of sensory neuron. In so far as we know this is not a question of the remodulation of a sensory neuron already present to accord with the locus of its receptor terminals, as suggested for the supernumerary limbs in birds and amphibians (Sperry, 1951; Székely & Szentagothai, 1962). Nor is it a question of the regeneration of fibres along a path that formerly carried a similar fibre type, as possibly occurs during regeneration of autotomized limbs in the Crustacea. The process involved in the formation of central connexions of fibres from the heteromorph is perhaps most comparable to that found in amphibia with functional, supernumerary eyes (Hibbard, 1959). Secondly, the regenerating fibres in the lobster enter and form connexions in an ‘adult’, functioning nervous system, not an embryological system as used in most vertebrate experiments. One must qualify this point, however, because loss by autotomy and subsequent regeneration of appendages is common in the Crustacea, so the distinction between an adult, differentiated system and an embryological, differentiating system may have relatively little meaning in lobsters. Thirdly, although first-order internuncials innervated by the regenerating afferents were not penetrated, and consequently no conclusions are possible with regard to modulation of these interneurons, yet there was apparently no change in the properties of the final motor neurons or in higher-order internuncials. Fourthly, unlike regeneration to the optic tectum in frogs (Gaze et al. 1963), neuron specificity was not evidenced primarily as geometrical localization, but rather as similar patterns of connexion and activity between specific kinds of neurons. Because of the extended arborizations of dendritic terminals in the lobster brain and the lack of appropriate anatomical information the actual importance of geometrical locus cannot be evaluated, but the ‘jig-saw-puzzle’ analogy of chemical specificity seems more apt for these experiments than does the ‘gradient hypothesis’ advocated by Gaze et al. (1963) for the frog optic tectum.

A single, well-studied individual, such as that described here, can tell us a great deal about what can occur in an organism. It gives no information, however, about the frequency of occurrence of the observed phenomena; and often, as in this case, very little about the time course and developmental stages of the processes. Such questions are currently under investigation in a series of lobsters. The full results will be reported at a later time, but two observations derived from these experiments are useful in evaluating the present report. First, several examples of functional heteromorphs capable of eliciting antennular responses very similar to those reported here have been found. The properties of this individual, therefore, do not necessarily represent a unique occurrence. Secondly, although most heteromorphs were functional, some were not, and some elicited only non-specific behaviour or abnormal antennular movements upon stimulation. The connexions described here, consequently, cannot be regarded as the inevitable result of eyestalk removal and heteromorph regeneration.

We wish to express our sincere appreciation to the Director and staff of the Bermuda Biological Station for their co-operation and hospitality. The lobster described in this paper was provided by the Government Aquarium in Bermuda through the generosity of its Director, Mr Louis B. Mowbray. We also wish to express our thanks to Mr R. B. Stephens, N.S.F. Undergraduate Research Participant, for his assistance in the experiments.

This work was supported in part by USPHS grants NB-03271 and NB-01624, and by a grant-in-aid to M. J. C. from the Bermuda Biological Station.

Fig. 3, Pl. 1. Response of heteromorph flagellum to anterior and posterior tipping. A. View from left side, normal position. B. Animal tipped up, flagellum moves down with eyestalk. C. Animal tipped down, flagellum moves up with eyestalk.

Fig. 9, Pl. 2. Responses of motor neurons to stimulation of heteromorph flagellum and antennular nerve bundle. A, Single electrical stimulus given at arrow to heteromorph flagellum. B, Single electrical stimulus given at arrow to either ipsilateral flagellar bundle (traces 1–4) or ipsilateral outer flagellum (traces 5 and 6). Each pair of traces : 1 and 2, 3 and 4, and 5 and 6 represent simultaneous dual recording. Trace 1, extracellular recording from ipsilateral antennular nerve; Trace 2, intracellular recording from unit 2; trace 3, extracellular recording from ipsilateral antennular nerve; trace 4, intracellular recording from unit 3; trace 5, extracellular recording from ipsilateral m. reductor,; trace 6, extracellular recording from contralateral m. reductor,. Calibrations, 40 mV.; 50 msec. Base-lines in traces 2 and 4 are drawn in.

Fig. 10. Pl. 3. Response of interneuron (unit 6) to antennular, heteromorph, and visual stimulation. Dual recording; upper trace in each pair from ipsilateral antennular nerve; lower trace, intracellular recording from interneuron. A, single, submaximal stimulus to ipsilateral flagellar bundle ; B, single, maximal stimulus to ipsilateral flagellar bundle ; C, repetitive (about 5/sec.) maximal stimuli to ipsilateral flagellar bundle; D, repetitive stimuli to heteromorph flagellum ; E, room light turned off at break in upper trace, interneuron responds with post-synaptic potentials after a brief latency. Calibrations, 10 mV., 100 msec. Base-lines are drawn in.

Fig. 11,Pl.4. Response ofintemeuron (unit 6) to repetitive stimuli. Dual recording upper trace in each pair from ipsilateral antennular nerve ; lower trace, intracellular recording from inter-neuron. Repetitive stimuli at about 10/sec. were given to: if, ipsilateral flagellar bundle; im, ipsilateral medial bundle ; cf, contralateral flagellar bundle ; cm, contralateral medial bundle ; or h, heteromorph flagellum. Stimulus artifact is obvious in recordings from antennular nerve. Calibrations, 20 mV., 100 msec.

Fig. 12, Pl. I. Response of interneuron (unit 10) to repetitive stimuli. Dual recordings; upper trace, ipsilateral antennular nerve; lower trace, intracellular recording from unit 10. Stimuli applied at about 1 o/sec. to : if, ipsilateral flagellar bundle ; im, ipsilateral medial bundle ; çf, contralateral flagellar bundle; cm, contralateral medial bundle; h, heteromorph flagellum. Calibrations, 40 mV., 100 msec.