Reflex Organization in the Swimmeret System of the Lobster: I. Intrasegmental Reflexes

The four pairs of abdominal swimmerets of the lobster perform rhythmic, metachronous beating movements which are used in larval and adult locomotion, righting responses, reproductive and perhaps respiratory behaviour (Davis, 1968a). Each swimmeret undergoes an alternating powerstroke (retraction) and returnstroke (protraction). During the powerstroke the two rami of the swimmeret spring open, increasing the exposed surface area of the appendage, and curl toward the rear, adding to the effectiveness of the powerstroke and positioning the rami for the returnstroke. During the returnstroke, the rami close and uncurl in preparation for the powerstroke of the next movement cycle (Davis, 1968a).

The motor output patterns which underlie these movements consist essentially of alternating bursts of impulses in two sets of antagonistic motor neurones (Davis, 1968b, 1969 a). This pattern of output occurs even in de-afferented preparations, showing that it does not depend upon timing cues provided by sensory feedback from the moving limb, but is instead endogenous to the CNS. Each swimmeret nevertheless possesses elaborate receptor systems which are capable of providing sensory feedback for the possible regulation of its movements. Indeed, elimination of these sensory systems disturbs the co-ordination of the swimmeret movements (Davis, 1968 c), indicating that the sensory feedback which they provide plays a significant role in shaping the motor output pattern.

The role of sensory feedback in regulating the movements of the swimmerets is the subject of the investigation reported in this series of papers. The present paper and the one which follows (Davis, 1969b) describe the intrasegmental swimmeret reflexes and their functional role in swimmeret beating. Subsequent papers will deal with the intersegmental reflexes (Davis, 1970 a) and the interactions between reflexive and endogenous central mechanisms in the swimmeret system (Davis, 1970b).

Eastern lobsters (Homarus americanus) were used. The abdomen was detached and the main flexor and extensor musculature was removed by dissection from the dorsal side, thereby exposing the ventral nerve cord and segmental nerve roots. The left half of the abdomen was secured in a chamber with the ventral surface down. The preparation was arranged so that the swimmerets were accessible from the side, and the unobstructed movement of each swimmeret in its natural arc was possible. The CNS was not damaged by the operation, and all sensory and motor connections between the abdominal ganglia and the swimmerets remained intact and fully functional.

During the experiments the preparations were submerged in oxygenated saline (Cole, 1941) which contained about 10 g./l. of granular dextrose. The saline was buffered with ‘TES’ (Sigma Chemical Co.) and maintained at 10−15°C. The reflex activity was especially sensitive to the pH of the saline, which was held at 7·2−7·4. With strict adherence to these procedures the preparations yielded useful data for several hours.

Glass pipette electrodes with fire-polished tips were attached by suction to various branches of the first abdominal nerve root to allow extracellular recording of efferent impulse traffic on its way to the swimmeret muscles. Swimmeret muscle potentials were recorded either extracellularly with suction electrodes on intracellularly with glass capillary micro-electrodes. Nerve and muscle potentials were amplified, displayed on a Tektronix oscilloscope and filmed with a Grass C-4 camera.

Controlled movements of the swimmeret were accomplished by means of the device illustrated in Text-fig. 1. The swimmeret was coupled to one end of a lever whose movement was monitored with a photocell system and taken as an indication of the position of the swimmeret. Stepwise movements of the swimmeret were obtained by moving the lever by hand. Sinusoidal movements of the swimmeret were obtained by coupling the lever to an eccentric point on a rotating wheel, which was in turn coupled through a gear box to a variable speed electric motor.

The swimmeret muscles, whose anatomy is described elsewhere (Davis, 1968b), are supplied exclusively by the ipsilateral first nerve root of the ganglion of the corresponding abdominal segment. The first root consists of two main bundles, anterior and posterior. At the base of the swimmeret the two bundles each divide into a dorsal and a ventral branch (Text-fig. 2). The anterior dorsal branch contains the efferent innervation of the main returnstroke (protractor) muscle (muscle 1−3 ; Davis, 1968b), while the posterior dorsal branch supplies the main powerstroke (retractor) muscle (muscle 4−8; Davis, 1968b). In the present experiments all of the nerve recordings were made from these two branches. The remaining muscles are contained within the swimmeret itself and are innervated mainly if not entirely by ventral branches of the posterior bundle of the first root.

The first nerve root is a mixed nerve, containing the swimmeret motor neurones as well as the sensory innervation of the swimmeret and parts of the lateral pleural plate. Most of the recordings were made from branches of the nerve which were presumably exclusively efferent, near their entry into the muscle which they supply. Sometimes, however, recordings were made from more proximal branches, which contain sensory as well as motor axons. To ensure that the reflex discharge recorded from these sites was motor rather than sensory, the following criteria were used, either alone or in various combinations: (1) rhythmic, spontaneous bursts of impulse similar to the efferent patterns which underlie swimmeret beating were recorded from the neurons in the absence of sensory feedback; (2) impulses in the neurones were correlated in a 1:1 fashion with potentials recorded from the swimmeret muscles; (3) the reflex responses of the neurones were abolished by severing the entire first root at its point of exit from the abdominal ganglion; and (4) stimulation of the sense organs which initiate the reflexes caused excitatory postsynaptic potentials and action potentials, both of which were recorded intracellularly from the somata of identified swimmeret motor neurones in the abdominal ganglia. The action potentials recorded from the somata were correlated in a 1:1 fashion with reflex spikes recorded peripherally from the motor neurone. In addition to these tests, the major reflexes reported here have also been demonstrated by recordings made directly from the swimmeret muscles rather than from the motor neurones (e.g. Text-fig. 13).

Impulses belonging to a single neurone were the same in amplitude and waveform. In these respects the variation between different motor neurones was usually large enough to allow unambiguous identification of single-unit activity.

Experiments have been performed on the swimmerets of all segments of both sexes. The intrasegmental swimmeret reflexes were qualitatively identical in all segments, and no systematic differences were found between the strength of the reflexes in different segments.

When a swimmeret was forcibly retracted and held, i.e. pushed into the position normally assumed at the end of the powerstroke, the excitor neurones supplying the main returnstroke muscle responded with an increased discharge rate (Text-figs. 3, 13). This reflex therefore resembles the ‘resistance’ reflexes found in the motor systems of many crustaceans (Bush, 1964, 1965; Bush & Roberts, 1968; Fields, 1966) as well as vertebrates (Granit, 1966). The reflex response showed an initial high-frequency phasic component, but decayed within several seconds to a steady, tonic discharge which was maintained for as long as the swimmeret was retracted (Text-fig. 4).

In some experiments all of the returnstroke motor neurones from which activity was recorded responded to retraction of the swimmeret with an increased efferent outflow. More often, however, a single axon in the returnstroke nerve responded to retraction of the swimmeret oppositely from the excitor neurones, i.e. its activity was reduced (Text-fig. 5). As expected, the phase histogram of the response of this neurone to sinusoidal movement of the swimmeret was usually shifted approximately 180° from that of the excitor neurones, i.e. its response approximately alternated with that of the excitors (e.g. Text-fig. 6). A similar reciprocity of activities occurred even in the absence of reflexive inputs, as shown by the following observations.

During some experiments the motor neurones were spontaneously active in patterns similar to those which normally occur during swimmeret beating. That is, rhythmic, alternating bursts of impulses occurred in the two main sets of antagonistic motor neurones. These patterns were produced before and after the tension response of the muscles deteriorated, and in all cases the swimmeret was immobilized. Thus, the production of the patterns was independent of the occurrence of sensory feedback, and undoubtedly reflected the activity of an endogenous oscillator in the CNS. During the production of such patterns, the same returnstroke neurone which showed reflex reciprocity with the excitors was active in bursts which alternated with bursts in the returnstroke excitor neurones (Text-fig. 7). Thus, the reciprocity between these units is also programmed into the central nervous motor score.

What is the function of the neurone which shows reciprocity with the excitors? The reciprocity alone suggests that the axon is a peripheral inhibitor. This interpretation is corroborated by experiments on the swimmeret system of the crayfish, where recordings from the main returnstroke nerve also show the alternation of simultaneous bursts in several motor neurones with a burst of impulses in a single neurone. The former are correlated with depolarizing potentials in the main returnstroke muscle, while the latter are associated in a 1:1 fashion with hyperpolarizing potentials (P. Stein, unpublished data).

The reciprocal effects of limb retraction on excitor and peripheral inhibitor axons could in principle result entirely from inhibitory connexions between these cells within the CNS. Assuming, however, that the results are indicative of true reflex reciprocity, the reduction of inhibitor activity by limb retraction could result either from active inhibition caused by sensory receptors responsive to retraction, or from the withdrawal of excitation from possible sensory receptors capable of signalling limb protraction. Additional experiments are needed to decide between these various interpretations.

When the swimmeret was forcibly retracted, the response of the motor neurones which supply the main powerstroke muscle was the same as the response of the returnstroke motor neurones, i.e. the retraction caused a phasic, high-frequency discharge which decayed within a few seconds to a steady, tonic discharge. The tonic response persisted for as long as the swimmeret was held in the retracted position (Text-figs. 8, 9). The reflex action of the powerstroke excitor neurones therefore further reinforces the movement which initiates it. Unlike the resistance reflex of the returnstroke neurones, the feedback to the powerstroke muscle is positive with respect to the movement.

As in the case of the returnstroke response, multi-unit recordings from the powerstroke nerve often contained a single unit whose response to retraction of the swimmeret was opposite from that of the excitor neurones (Text-figs, 10, 11). Impulses in this neurone were sometimes associated with small muscle potentials which were opposite in polarity from the muscle potentials affiliated with excitor neurone impulses. All of the available evidence thus suggests that this axon was a peripheral inhibitor, which also responds to reflexive inputs reciprocally with the excitor neurones supplying the same muscle.

The reflex responses to retraction of the swimmeret were not affected by removal of the distal three-quarters of the appendage. The sensory receptors which initiate the retraction reflexes are therefore located in the coxal region at the base of the swimmeret. Dissection of this region in fresh preparations revealed two elastic strands, earlier designated A and B (Davis, 1968b; Text-figs. 4, 5). These strands originate adjacently in the fork of the abdominal sternite, or rib, span the coxal joint and attach to movable portions of the limb by fan-like connective tissue insertions. Strand A inserts on the tendons at the insertion of the main returnstroke muscle, while strand B inserts on the anterior rim of the basipodite adjacent to the insertion of muscle 9 (Pl. 1).

Photomicrographs of sections of the coxal region show that each strand possesses an apparent nerve supply, presumably sensory in function (Pl. 1). The detailed structure and function of these organs has not been studied carefully, but the observations described here suggest that they are proprioceptors analogous to those found in other crustacean appendages (e.g. Alexandrowicz & Whitear, 1957; Bush, 1964; Burke, 1954; Cohen, 1963; Pilgrim, 1960; Taylor, 1967; Wiersma, 1959; Wyse & Maynard, 1965). These proprioceptors probably provide the sensory inputs for the retraction reflexes described above, although other sensory receptors, such as cuticular receptor cells in the membrane which articulates the swimmeret with the abdomen (Pabst & Kennedy, 1967), may also participate.

The edges of the two rami of each swimmeret are fringed with filamentous setae which represent a major source of sensory feedback for the swimmeret reflexes. These reflexes can be elicited by displacing the setae with jets of water. Streams directed from the front have little or no effect, but streams directed from the rear consistently elicit the reflexes. The setae therefore appear to be stimulated normally by the flow of water against them during the powerstroke.

The sensory axons from the setae converge in a large sensory nerve immediately before their entry into the basipodite of the swimmeret. Most of these axons are simultaneously excited by gently squeezing the rami distal to this point of convergence. This method of stimulation does not permit fine control of the stimulus parameters, but quantitative regulation of the stimulus was not required to demonstrate the reflexes.

Activation of the sensory axons from the setae by gently squeezing the rami with forceps elicited an intense burst of impulses in the excitor neurones which supply the main powerstroke muscle (Text-fig. 12). In contrast, the peripheral inhibitor axon to the powerstroke muscle responded to the same stimulus with a decrease in its discharge rate (Text-fig. 12). This reflex may also be considered a resistance reflex, since its action opposes the forward displacement of the swimmeret by water currents directed against the rami from the rear. During swimmeret beating, however, this reflex appears to provide positive feedback to the powerstroke muscle during the powerstroke, unlike the resistance reflex of the return-stroke excitor neurones in response to retraction.

Stimulation of the afferent pathways from the setae caused a strong central inhibition of the excitor neurones to the returnstroke muscle. Demonstration of this effect of course requires activity in the returnstroke excitor neurones. The most effective procedure is to retract the swimmeret, initiating discharge in the returnstroke excitors by way of input from the coxal proprioceptors. While the swimmeret is fixed in the retracted position the rami are then squeezed to inhibit the excitatory discharge to the returnstroke muscle (Text-figs. 13, 14).

Squeezing the rami vigorously often abolished even the most intense discharge in the returnstroke excitor neurones for several seconds. The duration of the inhibitory effect was considerably greater than the duration of the corresponding excitatory effect on the powerstroke excitor neurones (cf. Text-fig. 14 with Text-fig. 12). There-fore, although the method used here to elicit the setae reflexes was not the natural stimulus of water currents, the responses are nonetheless suggestive of a most effective negative feedback from the setae to the returnstroke muscle during the powerstroke.

In recordings from the main returnstroke nerve a single unit was sometimes seen whose response to stimulation of the setae was opposite from the response of the excitors, i.e. squeezing the rami increased its activity. This unit may have been the peripheral inhibitor axon which supplies the muscle, but independent evidence of its inhibitory function was not obtained.

Systematic recording from motor neurones other than those which innervate the main powerstroke and returnstroke muscles has not been attempted during activation of the swimmeret reflexes. Forcibly retracting the swimmeret often caused the rami to spring open and curl toward the rear, however, as they do during a voluntary powerstroke (see Introduction). If the tension response of the muscles had not deteriorated, the opening and curling movements elicited by imposed retraction appeared as vigorous as the same movements during swimmeret beating. These results suggest the possibility that the contraction of the corresponding muscles in the normal movement cycle may be controlled largely by reflexes activated during the powerstroke.

In an earlier study of the swimmeret movements it was found that the power produced by individual swimmerets is the same despite a probable rear-to-front decrease in the central ‘excitability’ of the segmental swimmeret oscillators. This result led to the hypothesis of a compensating reflexive amplification of the power output from the rear to the front of the abdomen (Davis, 1968 c). In the present work no systematic differences have been found between the strength of the intrasegmental swimmeret reflexes in different abdominal segments. A rigorous comparison of reflex strength of course requires good control over the strength of the stimulus used to elicit the reflexes. Although such control was lacking in the present experiments, the results nevertheless suggest that a segmental gradient in the strength of the intrasegmental reflexes can be reasonably excluded as the hypothetical rear-to-front amplifying mechanism. In the third paper of the present series, however, it will be shown that the intersegmental reflexes are capable of providing the proposed rear-to-front amplification (Davis, 1970 a).

The intrasegmental reflex pathways studied here are schematized in Text-figs. 15 and 16. The results provide a firm basis for the following interpretation of the reflex activity during swimmeret beating. During the powerstroke sensory inputs from the coxal proprioceptors and from the sensory setae which border the rami excite the powerstroke excitor neurones. The feedback from the movement is therefore positive. Input from the proprioceptors during the powerstroke also excites the returnstroke excitor neurones, but this excitation is completely negated by simultaneous inhibitory input from the sensory setae which border the rami of the swimmeret. Since these setae are stimulated by water currents only during the powerstroke, their inhibitory effect declines as the swimmeret decelerates in preparation for the returnstroke. The returnstroke excitors are thereby released from inhibition and are then able to respond to the excitatory input from the proprioceptors. The same excitatory input to the powerstroke motor neurones is at this time suppressed by the dominant central motor score (Davis, 1970b). The reflexive inputs to the main powerstroke and returnstroke muscles are organized to effect reciprocity of peripheral inhibitor axon activity with excitor neurone activity to the same muscle.

The above interpretation naturally requires the assumption that the reflexes can follow movements as rapid as those which occur normally during swimmeret beating. This is confirmed in the following paper (Davis, 1969b). From the above hypothetical reconstruction of events it is evident that reflex initiation of the returnstroke by the powerstroke movement is possible, but a converse initiation of the powerstroke by the returnstroke does not occur. In fact, no intrasegmental reflexes have been found which are capable of initiating the powerstroke. This point assumes special significance in the context of the discussion in the next section.

What role do the intrasegmental swimmeret reflexes play in generating the oscillatory motor output which underlies swimmeret beating? In the simplest case, reflex control of the cyclic pattern would be achieved by reciprocal negative feedback loops linking the powerstroke and the returnstroke movements. The necessary provision for reflex initiation of the returnstroke by the powerstroke has been demonstrated, but a similar provision for the initiation of the powerstroke by the returnstroke is lacking.

As emphasized above, no intrasegmental reflex has been found which is capable of initiating the power stroke. Anterior-going intersegmental reflexes linking the powerstrokes of different swimmerets have been identified (Davis, 1970 a), but these reflexes are incapable of initiating the powerstroke of the rear-most swimmeret. Moreover, a swimmeret performs normal movements even when input from more posterior segments is blocked. Therefore it appears that in the intact lobster the powerstroke of each movement cycle must be initiated anew by a purely central nervous motor command. Furthermore, all of the identified reflexive feedback to the powerstroke excitor neurones is positive, further reinforcing the powerstroke movement. Therefore it appears that in the intact lobster the cessation of the powerstroke discharge of each movement cycle is also controlled exclusively by the central motor score. The present results and the earlier work of others on the crayfish swimmeret system (Hughes & Wiersma, 1960; Ikeda & Wiersma, 1964; Wiersma & Ikeda, 1964) have shown that the CNS can generate the complex, oscillatory pattern of motor output which underlies swimmeret beating, even in the absence of rhythmic timing cues from sensory feedback. The present work has shown further that even when this sensory feedback is operative, the reflexes which it controls are in fact unable to directly initiate and/or sustain the cyclic output pattern.

The present study has shown that the intrasegmental swimmeret reflexes are organized so that they strengthen the powerstroke, and also reinforce the linkage between the powerstroke and the returnstroke within each cycle of movement. The reflexive inputs may also contribute to the reciprocity between excitor and peripheral inhibitor axon activities. The present work has shown, however, that each of these features of the motor output pattern is programmed into the central motor score independently of sensory feedback. Thus, the influence of the intrasegmental reflexes on the motor output pattern is quantitative rather than qualitative. In the event of an experimentally-induced conflict between the reflexive and endogenous central nervous mechanisms, the central command strongly dominates (Davis, 19706). The intrasegmental reflexes may therefore be viewed as relatively subservient amplifying mechanisms for independent neuronal oscillators in the CNS.

To what extent can the intrasegmental swimmeret reflexes automatically compensate for variable external loads, after the fashion of the vertebrate myotatic reflex (e.g. Granit, 1966)? As each swimmeret moves through the water during the powerstroke, it encounters a resistive force which is some function of the relative velocity of the swimmeret. Since the powerstroke velocity varies considerably in the adult (Davis, 1968 c) and probably even more so in the free-swimming lobster larva, the external load on the swimmeret undoubtedly normally varies over a large range. As the velocity of the swimmeret increases, the amount of sensory feedback from both the sensory setae and the coxal proprioceptors undoubtedly also increases. Any such increase is countered, however, by a simultaneous decrease in powerstroke velocity caused by the greater resistive force met by the swimmeret. True load compensation by the reflexes would thus require a velocity-dependent gain in the positive feedback loops which have been demonstrated. That is, as the velocity of the movement increases, the amplifying effect of the sensory feedback must increase faster than the corresponding resistance to the movement. Such a mechanism for automatically adjusting the motor output to handle variable external loads may be present in the swimmeret system, but if so, the method of load compensation is clearly based upon different principles than the follow-up length servo of the vertebrate myotatic reflex.

This work was supported by NIH postdoctoral fellowship NB 24, 882 to the author and by NIH grant 5 Roi NB 01624 to Dr Melvin J. Cohen. I am grateful to Dr D. Kennedy, Dr D. M. Wilson, Dr P. Stein, Dr A. Selverston and Dr D. Hartline for discussion and criticism of the manuscript.

(a) Section through the coxal region of the swimmeret, showing the elastic strand A (e.s.) at its point of insertion on the tendons (t.) at the base of the main returnstroke muscle, s.i., probable sensory innervation which terminates on the strand as a raised hillock (h.). The section was made parallel to the frontal plane of the lobster. b, c and d, sections parallel to the one in a, made through the coxal region of the swimmeret to show the elastic strand designated B (e.s.). The sections in b and d were made respectively at the mid-point of the strand and at its point of insertion adjacent to muscle 9 (b. 9; Davis, 1968b). The section in c was taken from a point between those in b and d. In c, s. i. designates the probable sensory innervation of the strand, while c.b. denotes possible cell bodies which resemble those found embedded in the connective tissue strands of many crustacean proprioceptors (e.g. Cohen, 1963).