Statocyst-Induced Eye Movements in the Crab Scylla Serrata: I. The Sensory Input from the Statocyst^*

The slow and fast phases of the eye movements of crabs during optokinetic nystagmus have been well investigated in intact animals, and a model of the whole system exists (Horridge & Sandeman, 1964; Barnes & Horridge, 1969). It has not yet been possible to extend these studies to include an electrophysiological investigation of the central nervous mechanisms because the optic lobes in the eyestalk do not withstand isolation although the isolated brains of crabs respond to the stimulation of sensory nerves (Sandeman, 1971). It has proved impractical to use visual stimuli to drive the oculomotor system in isolated preparations. Preliminary experiments, however, have shown that the oculomotor system in semi-isolated crab-brain preparations can be driven by irrigating the lumen of the statocyst with saline. The statocyst input alone is known to initiate nystagmus in crabs when they are rotated about the vertical axis (see review by Cohen & Dijkgraaf, 1961) and so our aim is to investigate fully the nervous control of crab eye movements that are generated by the directional input from the statocyst.

Before analysis of the central nervous integration can be made it is essential to know the sensory input in detail and we therefore begin by characterizing the responses of the statocyst sensory cells and describe the direction of fluid flow produced naturally by rotation of the organ about its vertical axis. In this paper we describe the anatomy of the horizontal and vertical canal systems in the statocyst and the extracellular responses of units in the statocyst nerves following irrigation of these canals.

Early comparisons of decapod statocysts with the vertebrate inner ear resulted in a belief that the statocyst was an organ of balance and hearing (Hensen, 1863). Subsequently the supposed auditory function of the organ was disproved, but it was also claimed that the crab statocyst could not function in the same way as the semicircular canals of vertebrates because the statocyst lumen is an open structure and not a system of canals (Prentiss, 1901), and this view persisted in the general literature. The similarity between the non-acoustic labyrinth system of vertebrates and the crab statocyst has nevertheless again been emphasized by Dijkgraaf (1955, 1956a, 1961).

The most complete early description of the receptor hairs in crab statocysts is that of Hensen (1863). He described three kinds of hair receptors within the statocyst-the hook hairs, group hairs and thread hairs. Dijkgraaf demonstrated, in Maja and Carcinus, the presence of a statolith supported on hook-shaped hairs. He called these ‘statolith’ hairs and renamed Hensen’s hook hairs as free hook hairs because they do not bear a statolith. Dijkgraaf showed that of the four types of hair, the slender thread hairs were the ones most likely to be associated with the detection of rotation of the animal about the vertical axis (Dijkgraaf, 1956a). There is no electrophysiological information on the responses of the various statocyst receptors in crabs although the responses of receptors in the lobster statocysts are known (Cohen, 1955, 1960).

The Australian mud crab, Scylla serrata, was used in all the experiments. Animals were kept alive in the laboratory wrapped in wet hessian.

The anterior part of the crab was separated from the rest of the body and the brain was removed after cutting the antennular nerves close to it. The proximal portion of the eyestalk was cut away to reveal the basal joint of the antennule. The rear wall of the basal joint was carefully cut out to expose the canal system of the statocyst.

Controlled irrigation of the statocyst lumen was achieved as follows. Two holes were made at each end of either the horizontal or vertical canal systems, depending on which group of receptors was under investigation. The tip of a glass pipette of internal diameter ±180 μm, was placed well into one of the holes and the intensity of positive or negative flow of saline through the tip was controlled by raising or lowering two reservoirs of saline attached to the pipette (Text-fig. 1). The level of saline covering the preparation, and the levels in the two reservoirs were compared in three vertical tubes. This allowed the establishment of a zero level and provided a means of calibrating the rate of flow. In some preparations the lumen of the statocyst was opened and the receptors were manipulated with needles. The onset and cessation of the flow through the pipette was monitored by a photocell attached to the stopcock which controlled the flow.

Electrical recordings were made after de-sheathing and splitting the various nerve bundles and lifting the strands on platinum-wire hook electrodes. A mineral-oil drop around the electrodes prevented the strands from drying out. Responses were tape-recorded and the recordings were then analysed and plotted with a Hewlett Packard 5480 B Signal Analyser to show the relative frequencies of the different receptors with respect to stimulus intensity.

The statocyst lumen of Scylla is contained in the basal joint of the antennule and is closed off from the outside by a suture along the dorsal surface of the basal joint. The lumen of the statocyst is shaped like two hollow toroids joined almost at right angles to each other (Text-fig. 2; Pl. 1, figs 1, 2).

The dorsal toroid, referred to here as the horizontal canal, is formed by a pinchingin of the upper and lower walls of the statocyst. Its circumference lies tilted about 15 ° from the horizontal, the medial edge being lower than the lateral edge.

The vertically oriented toroid, or vertical canal, is not completely joined at its centre but a central invagination of the medial wall of the statocyst (‘sensory cushion’ of Prentiss, 1901) comes within 80 μm of the opposite side of the statocyst, leaving a canal around it of 500–750 μm in diameter. The circumference of the vertical canal is tilted about 20° from the vertical and is also rotated approximately 45 ° about this axis so that its circumferential plane does not lie along the horizontal longitudinal axis of the animal. This is important with regard to the sensitivity of the system to displacement about both horizontal axes (see Discussion).

Within the canals there are four groups of hair-receptor organs which have essentially the same structure as the receptors described in Carcinus and Maja.

1. Large ‘group hairs’ are situated along the most lateral part of the horizontal canal (Text-fig. 3). They are 15 μm thick, 800 μm long, are not narrowed at their bases and are feathered along their length.

   Text-fig. 3.

   View largeDownload slide

   Dissection of the basal joint of the antennule and statocyst. The posterior walls of the horizontal and vertical canals have been removed and the positions of the group hairs, thread hairs, free hook hairs and statolith are shown. The thread hairs form a single line across the sensory cushion and project across the canals. The nerve supply to the free hook hairs (part of nerve II) has been removed to show the bundle of nerves running behind the statocyst to the bases of the thread hairs.

   Text-fig. 3.

   View largeDownload slide

   Dissection of the basal joint of the antennule and statocyst. The posterior walls of the horizontal and vertical canals have been removed and the positions of the group hairs, thread hairs, free hook hairs and statolith are shown. The thread hairs form a single line across the sensory cushion and project across the canals. The nerve supply to the free hook hairs (part of nerve II) has been removed to show the bundle of nerves running behind the statocyst to the bases of the thread hairs.

2. ‘Free hook hairs’ which lie along the posterior arm of the vertical canal (Textfig. 3). These are 40-60 μm an long, feathered, about 5 μm wide at their widest point and narrowed at their bases to 1–2 μm (Text-fig. 4). They are often bent into a hook like shape as their name implies.

   Text-fig. 4.

   View largeDownload slide

   Camera lucida drawings to show the disposition of the free hook hairs, statolith hairs and thread-hair bases within the statocyst. The free hook hairs are arranged along the posterior area of the vertical canal. Towards the bottom of the vertical canal the number of free hook hairs diminishes and there is a patch of statolith hairs surrounded by two circles of free hook hairs. The thread hairs are distributed over the sensory cushion in a single line, but there are proportionately more hairs at the dorsal end of the cushion than at the ventral end.

   Text-fig. 4.

   View largeDownload slide

   Camera lucida drawings to show the disposition of the free hook hairs, statolith hairs and thread-hair bases within the statocyst. The free hook hairs are arranged along the posterior area of the vertical canal. Towards the bottom of the vertical canal the number of free hook hairs diminishes and there is a patch of statolith hairs surrounded by two circles of free hook hairs. The thread hairs are distributed over the sensory cushion in a single line, but there are proportionately more hairs at the dorsal end of the cushion than at the ventral end.

3. ‘Statolith hairs’ are confined to a patch at the bottom of the vertical canal and surrounded by two arrays of free hook hairs (Text-fig. 4). They are simple unfeathered hairs 3–4 μm long and less than 1 μm in diameter.

4. ‘Thread hairs’ are arranged in a single line over the surface of the sensory cushion, but more hairs are concentrated on the upper part of the cushion and opposite the top of the vertical canal than at the bottom of it (Text-fig. 4). The hairs are about 300μm long and about 2 μm wide where they project into the canals and about one-third this length where the summit of the cushion lies close to the apposed wall of the statocyst.

The antennular nerve enters the brain as three main bundles (Text-figs. 2, 3). The most medial (I) extends from the distal joints of the antennule and does not receive any input from the statocyst. The two lateral bundles, II and III, contain sensory fibres from the statocyst and motor fibres which innervate the muscles of the basal joint of the antennule. The lateral nerves can be easily split into smaller bundles which come from different parts of the statocyst. Nerve II consists of three smaller bundles ; the two dorsal bundles contain fibres from the free hook hairs, and the ventral bundle contains only fibres from the thread hairs. Part of III carries sensory axons from the statolith hairs.

One bundle in the ventral part of II can be clearly traced into the sensory cushion, and electrical recordings from this nerve always give pure thread-hair-type responses (see below). Counts of thread hairs have shown between 80 and 96 hairs, but a count of fibres in the nerve bundle going into the sensory cushion reveals more than twice this number. The extra axons may be accounted for by there being more than one nerve cell for each receptor, as described for Astacus (Schöne & Steinbrecht, 1968).

Pure recordings from the thread hairs and statolith hairs were obtained relatively easily because most of the axons from these sense organs run in discrete bundles. Responses of free hook hairs alone were ensured by excising the statolith hairs and then, after recording, checking to see that all the statolith hairs had in fact been removed.

All the receptors can be made to respond, although with different stimulus intensities, by directing the saline current into a hole made in the lateral part of the horizontal canal and allowing the injected saline to escape through a hole at the medial end of the horizontal canal. Free hook hairs can be stimulated more directly by opening the top and bottom of the vertical canal and placing the pipette tip into the upper hole. Similarly the statolith hairs can be stimulated directly by opening the statocyst above the statolith and placing the pipette tip close to it.

Rotation of any circular canal containing a fluid from zero to a fixed velocity results in an initial movement of the fluid in a direction opposite to that of the imposed force. This movement continues until the frictional forces acting in the fluid and walls of the canal overcome the initial inertia of the fluid. The distance the fluid moves relative to the canal walls depends upon the viscosity of the fluid, the diameter of the canal, the acceleration of the canal and the terminal velocity of rotation. Relatively short pulses of saline applied to the canals therefore most closely resemble rotation of the statocysts, and all the receptors respond to this type of stimulus although with different sensitivities.

In this study we have applied not only short pulses of saline to the statocyst but also maintained stimuli. A maintained stream of saline does not occur naturally in the statocyst but can be used to stimulate static changes caused by gravity. Also, differences between the receptors can be more clearly brought out by choosing appropriate stimulus durations.

An indication of the relative intensities of stimuli applied while recording from the different receptor groups is given in the figures in millimetres. This is the difference in level between the stimulating reservoir and the level of saline in the preparation bath.

All single units were unidirectionally sensitive although there were apparently an equal number of receptors sensitive to deflexion in one direction or the other so that the statocyst as a whole does not appear to have a unidirectionally sensitive bias. The response characteristics of any particular group of hairs were the same, regardless of the direction of stimulation.

Direct stimulation of the statolith produces a response which can be maintained at a constant frequency for more than 60 sec. Characteristic of such a response is the relatively high-frequency initial phase which adapts rapidly (Text-fig. 5). The frequency of the following maintained discharge increases with an increase in the intensity of the flow of saline. Units always stopped firing abruptly when the stimulus was removed. Gentle mechanical pressure against the statolith always produced the same response as the water jet, provided the pressure was applied in the same direction.

The receptor discharge at the onset of the stimulus, as in the statolith receptors, is relatively higher in frequency than during the rest of the response, but the decline in frequency is less rapid. Units continue to fire for a few seconds after the stimulus is removed. Free hook hairs stop firing in response to a maintained stimulus after about 20 sec but unlike the thread hairs recover their excitability in a relatively short time (see below) (Text-fig. 6).

Thread hairs respond to the onset of the stimulus with a discharge which gradually increases in frequency as the hair is displaced. The rate of change of the spike frequency increases with increasing intensities of stimulus. A short stimulus pulse is characteristically followed by a long discharge. Direct observation of the thread hairs shows that the actual movement of the hairs is very slight and that they continue to respond while gradually returning to their original upright position. Like the thread hairs in lobsters, they probably have a set of discharge rate for any particular angular displacement from their upright position (Text-fig. 7).

Like the free hook hairs, the thread hairs also stop firing after 10 or 20 sec if they are subjected to a maintained current of saline. To recover from this treatment they need to be left undisturbed for about 3 or 4 min, but can be restored to full excitability within a much shorter time (30 sec) if the direction of the current flow is reversed (Text-fig. 8). This behaviour can be explained if it is assumed that, as in the lobster statocyst, the receptors respond only over the first few degrees of their movement from the vertical position and that beyond this point they are blocked (Cohen, 1960).

Comparisons between the sensitivities of the three different receptors following direct stimulation with the saline jet show that the thread hairs are the most sensitive, the free hook hairs next and the statolith hairs least sensitive (Text-figs. 5-7).

There is also a difference in the range of the responses of the thread hairs and free hook hairs. Thread hairs reach their maximum response frequency at a stimulus pressure of 2 mm whereas the free hook hairs and statolith hairs do not reach their maxima until stimulus pressures of 12 mm are applied.

The velocity of flow through the pipette tip can be calculated knowing the rate of flow and diameter of the pipette tip. The velocity of fluid flow in the canals, revealed by observing the passage of carmine particles through them, was about a fifth to a tenth the calculated velocity of the saline jet. Knowing the diameter of the horizontal canal we estimate that the threshold for the detection of rotation by the thread hairs is about 6° per second or 1 rev/min. Free-hook-hair responses to direct stimulation are almost as sensitive as those of the thread hairs but are much less sensitive to fluid flow directed through the horizontal canal. The free hook hairs would almost certainly be sensitive enough to detect rotation of fluid in the vertical canal and thus be capable of influencing eye movements following rotation about the horizontal axes, as has been suggested by Dijkgraaf (1956a).

Statolith hairs need a relatively high intensity, direct stimulus to make them fire, and normal irrigation of the horizontal canal did not excite these hairs. It must be pointed out that the isolated preparation is in the head-down position. The statolith receptors in this position may be less sensitive to fluid movement than when in their normal upright state. In general, irrigation of the canals with saline does not seem to be as effective a stimulus for the statolith hairs as manipulation of the basal joint of the antennule.

It is important for our investigation of the eye movements of crabs to know how closely the fluid movements in the statocyst caused by irrigation resemble those caused by rotation of the closed intact system.

Initially we considered the statocyst to approximate to two circular canals joined along part of their circumferences (Text-fig. 9 A). In this model the vertically oriented canal is inclined slightly out of the vertical, so that rotation about the vertical axis induces fluid movements in the same direction in both canal systems, although that in the horizontal canal will be more intense. The suggestion that this is not what occurs in the crab statocyst comes from observations of the movement of dye in a glass model, made to resemble the shape of the canals as closely as possible (Text-fig. 9B). Rotation of this model about its vertical axis showed clearly that the current in the vertical canal was in the opposite direction to that expected from the first model.

To test the model, the lumen of the statocyst was removed from the basal joint, suspended in a glass chamber of saline from the spindle of a small electric motor and observed from the side with a microscope. A small amount of dye was injected with a micrometer syringe and a glass micropipette through the translucent wall of the horizontal canal, near to its junction with the vertical canal. The micropipette was withdrawn from the canal and the statocyst was turned until the vertical canal was in view and then stopped. The resulting dye movement is slight but easily observable with rotational velocities of about 360°/sec. The movement of the dye confirms the prediction of the second glass model (Text-fig. 9B). Anticlockwise rotation of the right-side statocyst about its vertical axis causes a clockwise movement of the fluid in the horizontal canal including the part of the canal shared with the vertical canal. Fluid in the vertical canal moves down the posterior arm and up the anterior arm. The passage of dye also showed that the fluid movement in the vertical canal is confined mainly to the circumference and only very little flows over the centre surface of the sensory cushion (Text-fig. 9C).

Injection of dye into a statocyst with medial and lateral holes in the horizontal canal shows that most dye flows around the back of the canal, across the top of the vertical canal and also down through the lower arm of the vertical canal. The flow is therefore identical to that induced by rotation of the statocyst about the vertical axis.

The direction of fluid flow in the vertical canal revealed by dye experiments was confirmed using the directional responses of the free hook hairs as follows. The horizontal canal was opened medially and laterally, fluid flow was induced in both directions from the lateral hole and the responses of several free hook hairs were recorded. The pipette was then removed to the medial hole, the vertical canal was opened ventrally and the horizontal canal was blocked by injecting petroleum jelly into the lateral hole. Fluid movement was induced and the directional responses of the same free hook hairs (identified by amplitude) were recorded. The responses confirm that when positive pressure is applied to the lateral hole of the statocyst, fluid moves upwards in the posterior arm of the vertical canal (counter to arrows in Text-fig. 9C) and in the opposite direction with negative pressure (as depicted in Text-fig. 9C).

The movements of dye in the lumen of the statocyst indicate that in spite of the relatively open structure of the vertical canal fluid flow is confined mainly to its circumference so that the whole statocyst system can be compared with the vertebrate circular canals.

The dominant factor determining the direction of fluid flow in each canal is its angular relation to the axis of rotation. In the horizontal canals, for example, the most pronounced fluid movements occur during rotation about the vertical axis. Fluid in the vertical canal will be displaced by rotation about both the horizontal transverse and horizontal longitudinal axes because the plane of its circumference lies half way between these two axes (Text-fig. 10).

The second factor determining the direction of fluid flow is the shape of the two canals and the angles made at their junctions (Text-fig. 9 A, B). This is best demonstrated during rotation about the vertical axis when the greatest fluid flow will be induced in the horizontal canal. In our first model (Text-fig. 9 A) the horizontal and vertical canals join each other at an acute angle so that fluid is pushed past this junction to flow through the common arm and down the anterior arm of the vertical canal. In the statocyst, however, fluid at the junction of the horizontal canal with the medial canal is presented with a Y junction and can flow as easily down into the lower arm of the vertical canal as through the top of it, propelled in that direction by the rotational forces acting on it.

The third factor which may control the direction of fluid flow is the ratio of the resistances within the different canals. The effects of these differences are not easily measured on account of the complex shape of the statocyst. In cross-section all the canals have approximately the same diameter, but there are convolutions and constrictions in the canals which must modify fluid flow and which are not understood.

A table of the directions of fluid movements following displacement about the three main axes has been deduced from the model and used to predict the direction of thread hair deflexions and free hook hair deflexions (Table 1). The thread hairs are considered to form two main groups: the dorsal group in the common canal (THa) and the ventral group situated in the lower part of the vertical canal (THb) (Text-fig. 10).

Assuming that the directionality of the receptor responses is important, the table shows the information available to the central nervous system which is translated into an appropriate output to the eye muscles (Table 1). The difference between rotation about the vertical axis and the horizontal axes is clearly represented in the receptor output, because during rotation about the vertical axis the two groups of thread hairs are deflected in the same direction whereas during rotation about the horizontal axes the two groups of thread hairs are deflected in opposite directions with respect to each other. The movement of fluid around the lateral arm of the horizontal canal during rotation about the horizontal axes was determined from the model and shown to depart from the expected direction of flow as in the case of the vertical canal (Text-fig. 10). However, as there are neither thread hairs nor hook hairs in the lateral part of the horizontal canal, flow direction in this part of the system during rotation about the horizontal axes is unimportant.

One of the essential functions of the statocyst as a dynamic organ is to detect the direction of rotation about any axis. It is therefore interesting to find that as far as the free hook hairs and thread hairs are concerned, there are some horizontal axes which cannot be distinguished by one statocyst alone, as shown in Table 1. The right statocyst, for example, produces the same output for right-side-down and for headdown; also for left-side-down and for head-up. The left statocyst confuses right-sidedown with head-up and head-down with left-side-down. Table 2 compares the output from both statocysts and shows that where one statocyst confuses two stimuli the other discriminates between them. If the same conditions exist in the crab Maja, we can explain its ability to distinguish correctly between rotations about all axes with only thread hairs and free hook hairs intact (statolith input destroyed) as shown by Dijkgraaf (1956a, 1962).

In our records unit responses from different groups of hairs are remarkably homogeneous and, although within each group small differences of sensitivity occur, the three classes of responses are always clearly defined. The relevance of the small differences within each class is not known, but as single interneurones in the C.N.s. probably receive large numbers of sensory inputs, the small differences would immediately be lost.

All the unit responses were unidirectional, a property of many crustacean mechanoreceptors, and like other receptor systems (i.e. Pacinian corpuscles: Loewenstein & Mendelson, 1965; Ozeki & Sato, 1965; Hubbard, 1958; mammalian muscle spindles: Lippold, Nicholls and Redfearn, 1960) differences in the receptor responses are quite probably a reflexion of the mechanical compliance of the transducing structure (in this case the receptor hairs) and not any basic difference between the receptor cells. For example, one of the most striking differences in statocyst-receptor responses is revealed at the termination of a stimulus. Thread hairs continue to fire for a long period, free hook hairs for a few seconds and statolith hairs stop firing immediately. The thread hairs are very long and return very slowly to their original position and their responses reflect this gradual change of position. The free hook hairs are shorter and so return to their initial position more rapidly. The statolith hairs are even shorter, and because they are clustered together like the bristles of a brush, spring back into position immediately after they are released from the stimulus.

Whatever the reason for their differences it is clear that the impulse patterns of the hairs agree with the functions ascribed to them by Dijkgraaf (1956a, b), and the thread hairs are the receptors most likely to be of interest for the production of horizontal eye movements.

The function of the group hairs is not so obvious. They have been said to control the movement of the fluid in the canals (Dijkgraaf, 1961) and certainly they are supported on a very thin area of cuticle which if pressed in causes the hairs to swing out across the canal. How, or even if, this is achieved in the intact animal is not known. Stimulation of the motor nerves in the antennular nerve does not distort the canal system and produce any movement of the group hairs although contraction of all the muscles in the basal joint was observed ; so an active control system seems unlikely.

Stereopair photographs from the posterior (Fig. 1) and anterior (Fig. 2) of a statocyst dissected from the basal joint of the antennule. The toroidal shape of the horizontal canal is clear in both. In Fig. 1 the free hook hairs and statolith can be seen and in Fig. 2 the sensory cushion and group hairs are clearly visible.