Plasticity and Proprioception in Insects I

J. exp. Biol. 116, 435-461 (1985) 435 Printed in Great Britain © The Company of Biologists Limited 1985

PLASTICITY AND PROPRIOCEPTION IN INSECTS I. RESPONSES AND CELLULAR PROPERTIES OF INDIVIDUAL RECEPTORS OF THE LOCUST METATHORACIC FEMORAL CHORDOTONAL ORGAN

BY SASHA N. ZILL* Department of Biology, University of Oregon, Eugene, Oregon 97403, U.SA..

Accepted 12 October 1984

SUMMARY 1. The metathoracic femoral chordotonal organ is a joint angle receptor of the locust hindleg. It consists of 45—55 bipolar sensory neurones located distally in the femur and mechanically coupled to the tibia. 2. Responses of receptors of the organ were examined by extracellular and intracellular recording. The organ as a whole encodes the angle of the femoro- tibial joint but shows substantial hysteresis. Tonic activity is greatest at the extremes of joint position. 3. The organ possesses no direct linkage to tibial muscle fibres and shows no response to resisted muscle contractions in most ranges of joint angle. However, responses to extensor muscle contractions are obtained when the tibia is held in full flexion due to specializations of the femoro-tibial joint. These responses could be of importance in signalling preparedness for a jump. 4. Intracellular soma recordings of activity in individual receptors indicate that the organ contains two types of receptors: phasic units that respond to joint movement and tonic units that encode joint position and also show some response to movement. All units are directionally sensitive and respond only in limited ranges of joint angle. 5. Some phasic units increase firing frequency with increasing rate of movement and thus encode joint velocity. Other phasic units fire only single action potentials and can encode only the occurrence and direction of joint movement. All tonic units increase activity in the extremes of joint position and show substantial hysteresis upon return to more median positions. 6. Direct soma depolarization produces different responses in different types of units: phasic receptors show only transient discharges to current injection; tonic receptors exhibit sustained increases in activity that are followed by periods of inhibition of background firing upon cessation of current injection. 7. Receptors of the chordotonal organ are separable into two major groups, based upon their response characteristics, soma location and dendritic orientation: a dorsal group of receptors contains tonic units that respond in ranges of joint flexion (joint angle 0-80°) and phasic units that respond to

•Present address: Department of Anatomy, University of Colorado Medical School, Denver, Colorado 80262, U.S.A. Key words: Joint angle receptors, proprioception, cellular physiology. 436 S. N. ZILL flexion movements; a ventral group of sensilla contains tonic units active in ranges of joint extension (joint angle 80-170°) and phasic receptors that respond to extension movements. 8. The response properties of these receptors are discussed with reference to the potential functions of the chordotonal organ in the locust's behavioural repertoire.

INTRODUCTION Many sense organs, in both vertebrates and invertebrates, precisely monitor the angles of joints of appendages (Granit, 1955; Mill, 1976). While the responses of joint angle receptors of vertebrates have been studied in detail (Boyd & Roberts, 1953; Ferrell, 1977), little is known about how information provided by these receptors is incorporated into behaviour (Grigg, Harrigan & Fogarty, 1978; Baxendale & Ferrell, 1981). In contrast, recent work in invertebrates has shown that during certain behaviour, such as walking (Graham & Bassler, 1981) and jumping (Steeves & Pearson, 1982), input from joint angle receptors can extensively modify posture and movement. However, other more stereotyped behaviour, such as stridulation (Bassler, 1979) and flight (Wilson, 1961), can be performed in the absence of inputs from joint angle receptors or with this input experimentally disrupted. Despite the advantageous accessibility of invertebrate nervous systems for study at a cellular level (Hoyle & Burrows, 1973), the neuronal mechanisms underlying this implied plasticity of behavioural effects of joint angle receptors have not yet been determined. The present series of investigations, therefore, were undertaken to study a single, identified group of joint angle receptors of the locust hindleg, the metathoracic femoral chordotonal organ. The goals of these investigations are to define the properties of sensory and motor elements of the locust nervous system that determine the functions of this group of receptors and permit flexibility of coupling in behaviour. The locust metathoracic femoral chordotonal organ was first described by Usherwood, Runion & Campbell (1968), who examined the morphology and responses of the organ and identified it as a joint angle receptor. However, a number of basic questions remained as to the responses of the organ. As noted by Burrows & Horridge (1974, p.59): 'No doubt many motorneuron responses to joint motion are due to the chordotonal organ, but inferences about its central action are limited because we lack the following details: (a) whether some or all units are directional in their response, (b) whether the two directions of motion excite the same or different units in different parts of the range, (c) whether vibration at different parts of the range excites different units which could thus signal position although not tonically.' Further, a number of experiments by Bassler (1968), utilizing lesion or disruption of afferent input, have shown that the organ can substantially affect walking and jumping. However, the interpretation of some of these experiments has been confused by a lack of knowledge about the effects of these operations on chordotonal organ output. For example, Bassler (1968) showed that after cutting one of the ligaments of the organ, jumping could not be elicited. Heitler & Burrows (1977) attributed this Insect proprioception 437 effect, not to a discrete function of the chordotonal organ in the jump, but rather to the fact that the locust perceived its joint as fully extended. In contrast, Pearson, Heitler & Steeves (1980) postulated that the chordotonal organ provided inputs that specifically triggered jumping, although the mechanism by which a joint angle receptor could trigger a movement in a joint held immobile by the co-contraction of antagonist muscles was not determined. The first paper in this series, therefore, re-examines the basic morphology and responses of the chordotonal organ, and also studies the responses of individual receptors by intracellular recording. Subsequent papers utilize the information provided by this study to investigate the specific effects of the chordotonal organ upon motoneurones and interneurones. Few previous studies have examined the responses of chordotonal sensilla by intracellular recording (Mendelson, 1963, 1966), owing in part to technical problems. The locust femoral chordotonal organ has proved amenable to such studies. The cellular properties of these receptors may provide insight into the functions of joint angle receptors in behaviour.

METHODS Adult male locusts {Schistocerca gregaria), provided by the University of British Columbia and maintained in laboratory cages at 25°C, were used in all experiments.

Anatomy Animals (N = 7) were induced to autotomize one metathoracic leg that was pinned out in a Sylgard resin-coated dish, with the femoro-tibial joint angle at 80°. After removing the tarsus, the leg was gently perfused with Karnovsky's fixativefo r 30 min. The parts of the distal femur and proximal tibia containing the femoral chordotonal organ and its attachments were then cut out, dehydrated and embedded in Spurr's resin. Serial sections (2 fim) were taken either perpendicular or parallel to the long axis of the leg, stained with toluidine blue, and examined by light microscopy. Individual sections were traced through a drawing tube onto acetate sheets and composite overlays were constructed.

Physiology Extracellular recordings Intact animals were restrained in wax so that the outer surface of the femur was held in a horizontal plane facing upwards. A small window was cut into the distal femur to expose the main ligament and flexorattachmen t of the organ and to cut nerves distal to the receptors. Another window was cut 8-10 mm proximal to the organ and multi- unit recordings were taken from the whole nerve 5bl (Campbell, 1961) using hook electrodes. The femoro-tibial joint angle was monitored either by a potentiometer attached to the tibia (Young, 1970) or with a capacitative transducer (Sandeman, 438 S. N. ZILL 1968). Movements of the tibia were generated manually or with a piezoelectric crystal (as described below). All data were stored on tape for subsequent analysis.

Intracellular recordings Animals were mounted dorsal side up and the metathoracic legs held in wax so that the inner surface of one femur was in a horizontal plane (Fig. 1). An insect pin, whose ends had been bent into small loops, was glued to the proximal tibia. One of these loops served as a swivel joint when linked to another pin that was attached to the end of a piezoelectric crystal. The crystal generated smooth movements of the tibia when driven by voltages derived from a wave-form generator. Movements were monitored by a photoelectric cell placed close to a flag that was attached to the tibia. To expose the femoral chordotonal organ, a small window was cut in the cuticle of the distal femur. The tibia was then fully extended and a section of the tendon of the extensor tibiae muscle was removed. The joint was then moved to an angle of 45 ° and a

Intracellular electrode Femur PE crystal

Fig. 1. Diagram of preparation for intracellular recording. A locust is mounted in wax so that one metathoracic leg lies in a horizontal plane with its inner (medial) surface upwards. Changes in tibial position are generated by a piezoelectric crystal (PE crystal) linked to the tibia by a small pin. Displacements are monitored by a photocell placed close to a flagattache d to the distal tibia. A small window is cut in the distal femur to expose the femoral chordotonal organ (CO, enlarged for clarity). The organ is stabilized by a small platform and penetrated with microelectrodes. Insect proprioception 439 portion of the flexor tibiae tendon proximal to the attachment of the chordotonal organ was excised. This dissection exposed the main body of the chordotonal organ and effectively left the tracheal supply of the organ intact as it derives from the outer surface of the organ. The presence of an intact tracheal supply and its aeration by the animal proved to be essential for prolonged recordings from the chordotonal organ. In early experiments, performed on isolated legs or those involving ventral dissection of the leg and disruption of the tracheal supply, preparations rapidly deteriorated as judged from decline in resting potential and poor mechanical response. In contrast, recordings from dorsally dissected organs showed stable resting potentials and consistent responses could be elicited from single receptors for up to 1 h. A small amount of saline was placed over the preparation to prevent drying and to link it to a ground wire placed close to the opening. The saline used was kept to a minimum and was seen to mix rapidly with the animal's own circulating haemolymph. A small platform was manoeuvred under the proximal portions of the organ for mechanical support. Single bipolar neurones of the organ were then penetrated with glass microelectrodes filled with either ZmolP1 potassium acetate (50— 70 MQ) or a 5 % solution of Lucifer Yellow (kindly provided by Dr Walter Stewart). For dye- filling, negative current pulses of 1-3/iA were applied for 500 ms at 1 Hz. Chordotonal organs were then fixed, cleared and examined conventionally. In other experiments, intracellular recordings were taken from axons of chordotonal sensilla in nerve 5 close to the metathoracic ganglion. The ganglion was exposed, supported and perfused with saline according to the method of Hoyle & Burrows (1973). Individual receptors were identified by their short latency responses to lifting of the main or flexor ligaments of the organ (exposed as above) by means of a small hook attached to the piezoelectric crystal.

RESULTS Anatomy The femoral chordotonal organ is composed of 45-55 (mean 48-8, N =1) bipolar neurones located in the distal femur 6-7 mm proximal to the femoro-tibial joint. Axons of these neurones join nerve 5bl (Campbell, 1961) by a short branch to project to the central nervous system. The organ is firmly attached to the outer (anterior) wall of the femur by a short attachment ligament (Fig. 2). In light micrographs, two structures were consistently found that couple the receptors to the next distal leg segment, the tibia: a main ligament that terminates adjacent to the insertion of the tendon of the extensor tibiae muscle, and a flexor ligament that attaches directly to the tendon of the flexor tibiae muscle, approximately 5 mm proximal to the femoro-tibial joint. While Usherwood et al. (1968) state that several other small ligaments link the organ directly to fibres of the extensor and flexor muscles (based upon examination of whole mount preparations) these ligaments could not be identified in seven preparations serially sectioned and examined by light microscopy in the present study. A small nerve was regularly found to continue distally from the organ and branch in the 440 S. N. ZILL

; *' 0®fr *

Fig. 2. Structures of the chordotonal organ and individual receptors. (A) Composite light micrograph showing neurones and ligaments of the chordotonal organ. The organ is composed of two groups of bipolar neurones (dorsal and ventral) whose axons join nerve Sbl (n5bl). An attachment ligament (a) anchors the organ to the cuticle (c) of the femur. Two ligaments link the organ to the tibia: a main ligament (m) and a flexor ligament (/"). A small nerve proceeds distally from the dorsal surface of the organ to innervate hairs of the distal femur. (B) Individual chordotonal sensilla (filled with Lucifer Yellow) differ in their dendritic orientation. Dendrites of receptors of the dorsal group (left) are orientated parallel to the long axis of the femur (arrow); receptor dendrites of the ventral group (right) are angled with respect to the same axis. Calibration bar: A, 100^m; B, 75 ftm. region of the extensor muscle. This nerve, which contains only axons from small hairs located on the distal femur (as indicated by extracellular recordings) could have been mistaken for a ligamentous attachment. Thus the only mechanical coupling of the organ appears to be by two ligaments that link it to the tibia. Movement of the tibia produces selective stretching of the organ by each of these ligaments that is readily Insect proprioception 441 discernible in dissected preparations (Fig. 3A): decreasing joint angle in flexion progressively stretches the organ by the main ligament; increasing joint angle in extension stretches the organ by the flexor ligament. Receptor cell somata and scolopale terminations of receptor dendrites in individual sections were drawn on acetate sheets (by means of a drawing tube) and composite overlays constructed from serial sections. Several features were noted in each of the seven composites constructed. First, the cell bodies of the receptors often formed discrete dorsal and ventral groups. Within these groups the largest somata were

EXTEND t

• Tibia

Attachment Axis of femur

Main ligament I

Fig. 3. (A) Diagram of insertions of chordotonal ligaments. The femur and tibia are shown in outline. The joint is hinged and can move only in flexion or extension (in the plane of the diagram). Two antagonist muscles produce these movements, the extensor and flexor tibiae (their tendons are drawn as dark lines). The chordotonal organ (enlarged) has two ligaments that link it to the tibia: the main ligament is stretched during joint flexion, the flexor ligament is stretched during joint extension. (B) Orientations of dendritic insertions in the organ. A composite diagram, constructed from acetate overlays of drawing of serial sections of one chordotonal organ, indicates the orientations of the dendritic terminations (scolopales): scolopales of dorsal neurones are orientated parallel to the long axis of the femur; scolopales of ventral sensilla are angled with respect to the femoral axis. 442 S. N. ZILL located proximally, smaller cells generally were found more distally. Often the axons of the dorsal and ventral groups formed separate bundles that joined nerve 5b 1 individually. Second, the scolopales of the receptors showed two distinct orientations (Fig. 3B): scolopales located dorsally were orientated parallel to the long axis of the leg, while ventral scolopales were angled (30-35° with the leg fixed at 90°) with respect to the same axis. These observations suggest that, anatomically, the femoral chordotonal organ is composed of two discrete groups of receptors.

Physiology Extracellular recordings Tonic discharge and hysteresis. Tonic activity of the chordotonal organ was examined by extracellular recording from nerve 5b 1, with all nerve branches distal to the organ cut while the femoro-tibial joint was set at different angles. The upper plot in Fig. 4A shows the tonic activity of the organ when the tibia was moved away from the mid-position (80°) progressively into extension (160°) and flexion (0°). The organ as a whole shows a corresponding increase of activity in each range away from the rest position. However, this discharge showed considerable hysteresis and spiking activity depended upon the direction from which a particular joint angle was approached. When the tibia was moved back toward the mid-range of joint angle a sudden decrease in firing rate occurred that remained depressed for periods up to 1 min (Fig. 4A, lower trace). This hysteresis depended both upon the magnitude of joint angle traversed in return and the specific range of angle, being greater in joint flexion than extension. Similar hysteresis has been noted in other chordotonal organs (Burns, 1974) and other joint receptors in insects (Coillot & Boistel, 1969). Contributions of different ligaments to tonic response. To evaluate the effects of the two insertions of the chordotonal organ, each ligament was individually cut in the distal femur and discharge at different joint angles was recorded extracellularly (Fig. 4B). Severing the main ligament eliminated the increase in firing rate in joint flexion but did not affect the discharge to joint extension. Cutting the flexor ligament produced a discharge upon joint flexion that was unchanged and almost completely eliminated the increment of response in joint extension. A small increment occurred in the most extreme range of joint extension. It should be noted that after these ablations the chordotonal organ discharge in the affected range did not fall to zero but remained approximately equivalent to that seen at 80°. Phasic discharges. To examine the responses of the chordotonal organ to joint movement recordings were taken from n5bl while the tibia was moved at different rates through angles of 10-15° using sinusoidal or ramp waveform inputs to the piezoelectric crystal. Fig. 5 shows the results of one experiment using sinusoidal frequencies of 0-2, 0-5 and 5 Hz over the range of 90-105° of joint extension. Several characteristics of the response to joint movement can be noted. First, discrete discharges occur during rapid joint movement. Inspection of the second and third traces in Fig. 5 clearly shows that units of different size respond to each direction of movement. Similar discharges were observed in all ranges of joint angle. Thus, the Insect proprioception 443

120

100

3 CT" 0 20 40 60 80 100 120 140 160

120 B

100 \ A 80 \ A' 60 \ A 40 V / 20

0 20 40 60 80 100 120 140 160 Joint angle (degrees)

Fig. 4. (A) Tonic responses of the chordotonal organ. Activity of the organ was recorded extracellularly while the tibia was moved from an angle of 80° in 20° steps progressively into flexion and extension. All action potentials above baseline were counted. The graph plots the discharge frequency (+S.D.) for 20 consecutive seconds, 30 s after tibial displacement. Discharges increase as the tibia is moved away from the median range. Return movements show substantial hysteresis and much lower firing rates are attained. (B) Contribution of different ligaments to tonic discharge. One of the ligaments of the organ was cut in each of two animals and discharge of the organ was tested as above. Severing the main ligament (dashed line) eliminates the increase in firing rate in ranges of joint flexion but leaves intact the discharge to joint extension. Cutting the flexor ligament (solid line) produces normal responses to joint flexion but almost completely eliminates the increase of activity in joint extension. chordotonal organ appears to contain phasic units that respond to any movement at any starting position. Secondly, units in the size range of those that were tonically active at rest (top trace) showed a modulation of their frequency during joint movement, increasing in movements away from 80° (joint extension) and decreasing upon return (joint flexion). At higher frequencies, complete inhibition of tonic activity occurred during return (flexion) movements. Similar inhibition was noted following step displacements, corresponding to the hysteresis previously described. 444 S. N. ZlLL Responses to vibration. To examine the potential responses of the femoral chordotonal organ to transmitted vibration, extracellular recordings were taken from

Extend 0-2 Hz

Stim.

0-5 Hz

5 Hz

Fig. 5. Discharges during joint movement. With the tibia at rest (joint angle 110°) irregular firing of tonic units is seen. During sinusoidal movements (15° at 0p2, 0-5, 5-0Hz) phasic units are active. Units with different sized spikes fire in response to joint flexion or joint extension. Activity of tonic units is inhibited by joint flexion (up on position trace) and increases upon joint extension. In very rapid movements tonic activity is completely inhibited during flexion movements. Calibration: first trace, 0-3 s; second trace, 1-25 s; third, fourth trace, 0-1 s. CO, chordotonal organ. Insect proprioception 445 the organ with the femoro-tibial joint completely immobilized with epoxy resin while sinusoidal mechanical stimuli were applied at the distal tibia with a piezoelectric crystal. No responses were obtained in these preparations in the absence of joint movement. In other experiments, where the joint was free to move, vibratory stimuli were applied by tapping on the experimental mounting or by placing a variable speed motor adjacent to the preparation. Again, no consistent responses occurred without visible joint movement. Although some frequencies of vibration may be transmitted by the organ in freely moving animals, sensitivity is probably severely limited by viscosity of the tibial muscles. A lack of responsiveness to vibration has also been noted for other joint chordotonal organs (Hustert, 1982). In general, chordotonal organs that respond to vibration do not span leg joints (Schnorbus, 1971). Response to muscle contractions. A number of chordotonal organs have been shown to be directly mechanically linked to leg muscles (Burns, 1974; Clarac, 1968). Usherwood et al. (1968) also speculated that the metathoracic femoral chordotonal organ responded to resisted muscle contractions. In the present series of experiments activity was recorded from the organ during spontaneous movements by the animal and during stimulation of the nerves to the extensor and flexor tibiae muscles. Chordotonal organ activity during spontaneous movements (Fig. 6Ai,ii) was similar to that seen during imposed displacements and there was no indication of a resetting of afferent activity by muscle contractions, as occurs in vertebrate muscle spindles (Matthews, 1972). Also, with the joint completely immobilized, no change of activity was recorded during stimulation of either the extensor or flexor nerves (with one exception, described below). Thus, through most of the range of joint angle there is no direct efferent control of chordotonal organ activity. A change in chordotonal activity was produced by extensor muscle contractions when the joint was placed in full flexion and movement of the tibia blocked by a small pin or held by spontaneous flexor activity. Repeated twitch contractions of the extensor muscle produced an increase in chordotonal organ activity (Fig. 6B). This firing was not due to a direct effect of fast extensor activity upon the organ but rather to the effect of extensor contractions upon the femoro-tibial joint. Resisted contractions of the extensor muscle at full joint flexion have been shown to produce bending of the distal end of the femur, storing energy for the jump (Bennet-Clark, 1975; Heitler, 1977). These contractions also result in a bending of the proximal end of the tibia and displacement of the insertion of the main ligament of the organ (50-60/xm as measured by an optical micrometer). The resultant discharge is of importance as the chordotonal organ can thus monitor the extent of tibial bending and signal preparedness for the jump.

Intracellular recordings Individual chordotonal sensilla regularly showed resting potentials of 50—65 mV in intracellular recordings. The most stable penetrations were obtained from the proximal portions of the organ, that is, the region containing the receptor cell bodies. In this area, overshooting action potentials (70-90 mV in amplitude) could regularly 446 S. N. ZILL

Ai Joint angle Extend

co y

Fig. 6. Chordotonal organ activity (CO, lower traces) recorded extracellularly during spontaneous movements of the tibia. The femoro-tibial angle (upper traces) was monitored by a Sandeman transducer. Movements generated by the animal produced activity similar to that seen during imposed changes in joint angle. (Ai) A large joint flexion (80°-0°) produces phasic activity during the movement and sustained firing when the tibia is held flexed, (ii) A joint flexion is followed by a large joint extension (80°-140°). Return to the initial joint angle is accompanied by hysteresis and slow recovery. (B) Activity during spontaneous repetitive extensor contractions. With the tibia held in full flexion and joint movement prevented, a burst of activity in the fast tibial extensor motoneurone (large spikes in tibial myogram, TM, lower trace) produces intense afferent activity that subsides after motoneurone activity ceases. Calibration: A, 500 ms; B, 100ms.

be recorded. Cells in different parts of this proximal region showed responses in different ranges of joint angle (see below). The entire angular range of response of a single receptor could not be examined in soma recordings since large changes in joint angle produced a stretch of the organ and a shift in soma position sufficient to pull the Insect proprioception 447 recording electrode out of the cell. Consequently, responses could be routinely examined in soma recordings only through arcs of 10-15°. Some receptors were examined over their full response ranges through axon recordings (see below). However, single preparations were studied in several different ranges of joint angle and cell populations that did not show mechanical responses at some joint angles were active in others. The results presented below represent recordings from 72 cells in 18 preparations with as many as nine cells recorded from a single organ. Phasic units. Many cells (N = 49) did not exhibit activity at rest but showed vigorous phasic responses to joint movements (Fig. 7). Phasic units were always directionally sensitive, responding to either extension or flexion movements but not

A Stim.

-ill J L

Ei Eii

Fi Fii

Fig. 7. Phasic unit responses. Individual phasic receptors are directionally sensitive in their responses to joint movement firing to extension (A) or flexion (B) but not both directions. Some receptors produce multiple spikes to joint movements (C) while others fire only single action potentials at all frequencies of oscillation (D). Many phasic units show increases in firing frequency and decreases of burst duration during increasing frequencies of sinusoidal (E) and ramp (F) movements. Calibration: vertical displacement A,B, 25°; C,D, 20°; E,F, 10°; voltage: 90mV; horizontal A,B,E,0-5s; C,D, 0-1 s;F, 0-2s. 448 S. N. ZILL both. However, in all ranges of joint angle different individual receptors could be located that responded to either movement, so the organ as a whole detects joint displacements in any direction in agreement with observations based on extracellular recordings. The responses of individual phasic units depended upon their position within the organ. Cells responding to flexion movements were found in the dorsal part of the organ, while cells sensitive to extension movements were located ventrally, close to the origin of the flexor ligament. Examination of phasic units filled with Lucifer Yellow showed that receptors with different directional sensitivities were also morphologically distinct in their dendritic orientation (Fig. 2B). Cells of the dorsal group had dendrites that inserted nearly parallel to the long axis of the leg. Receptors of the ventral group had dendrites that were angled (25-35°) with respect to the leg axis and inserted close to the origin of the flexor ligament. Thus the responses of individual phasic receptors were correlated with both their position and their dendritic orientation. Somata of phasic receptors were often quite large (maximum diameter llO^m) and their dendrites quite long (maximum 170/im). Some phasic units (N— 14) showed only single spikes to all frequencies of joint movement (Fig. 7A,B,D) and thus function solely as movement detectors. These units were often, however, extremely sensitive and spiking could be elicited by joint movements as small as 0-3/im (approximately l-2minutes of an arc). Other units (N = 14) (Fig. 7C,E,F) fired repetitively to single movements and the frequency of firing depended upon the velocity of joint movement. Fig. 8 is a plot of the responses of two different phasic units, one extension and one flexion sensitive, to alternating movements of 12° at different frequencies. At low velocities of joint movement, these units respond to increasing rate of joint movement with a linear increase in firing frequency. At frequencies greater than 50—75 Hz these units show response saturation. Similar response saturation has been found in phasic receptors of other chordotonal organs (Bush, 1965). Tonic units. Many units in the chordotonal organ (TV = 23) showed tonic activity. This firing was always irregular over time and patterned activity or regular grouping of spikes did not occur (Fig. 9Ai). All tonic units increased their activity when the tibia was moved away from 80°, i.e. to increasing flexion or extension (Fig. 9Aii). Individual units, however, were directionally sensitive, increasing activity to either extension or flexion but not both. All tonic units also showed some phasic properties: small-to-moderate step displacements produced an initial high firing rate that adapted, with some irregularity over time (Fig. 9Aiii,iv). Step returns to the initial tibial position were often accompanied by an inhibition of background activity. This was particularly apparent in large displacements where firing could be completely inhibited for several seconds and full recovery could take up to 1 min (Fig. 9Av). This inhibition paralleled the hysteresis shown to step displacements by the organ as a whole, although unit responses to large displacements producing prolonged inhibition could not be studied in soma recordings. Tonic units also showed directionally sensitive modulation of their firing frequency during imposed movements of the tibia. Slow ramp movements resulted in irregular increases in firing rate as the Insect proprioception 449

u 3 40

I 30

10 •

0 3 4 5 6 7 10 Stimulus frequency (Hz)

Fig. 8. Responses of phasic units at different velocities of joint movement. The discharge frequencies (ordinate) of two units, one flexion sensitive (upper), the other extension sensitive (lower) are plotted at different velocities of joint movement (abscissa) over an arc of 12° starting at a joint angle of 30°. Both units show linear increases in firing frequency at low velocities of joint movement but saturation at higher velocities. Each point is mean discharge + S.D. over 25 consecutive cycles. tibia was moved away from the mid-range of joint angle and decreases upon return (Fig. 9B). During rapid ramp or sinusoidal movements complete patterning of activity occurred with spiking during movements in one direction, and complete inhibition upon return (Fig. 9C). The majority of tonic units were identical to phasic units in their directionally sensitive distribution within the organ and their dendritic orientation. Thus, units tonically active in joint flexion were found in the dorsal group of receptors and had dendrites orientated parallel to the leg axis. Units active in joint extension were found in the ventrally located group and had dendrites that were angled with respect to the leg axis. Three units however were noted to be exceptional. These units were active in extreme extension (140-160°) and were penetrated close to the attachment of the organ. Units close to the attachment were extremely difficult to penetrate and hold in intracellular recordings; however, a Lucifer fill of one such unit showed a dendritic orientation characteristic of the dorsal group. Lucifer fills of tonic units showed cell bodies that were always smaller than those seen in phasic units (maximum diameter, 60 /im) and often had substantially shorter dendrites (maximum length, 100/im). 450 S. N. ZILL Effects of direct displacement of ligaments upon unit responses. To further evaluate the relative contributions of the two major insertions of the chordotonal organ, responses of individual receptors were examined to both displacement of the tibia and lifting or pulling of the ligaments. Units of the dorsal group that responded to joint flexion also fired in response to stretch of the main ligament. Receptors of the ventral group that fired in response to joint extension also responded to stretch of the flexor ligament. The responses of some phasic units in each group were exceptional. In ranges of joint extension, some phasic receptors of the dorsal group, that showed activity in response to joint flexion and to stretch of the main ligament, also responded to release of stretch of the flexor ligament. Other receptors were also located in the ventral group that responded to both joint flexions and release of the main ligament.

Stim.

CO- JUUUUL

Fig. 9. Tonic unit responses. Individual tonic units show irregular baseline levels of activity that depend upon joint position, increasing as the joint is moved away from the median range (Ai, 40°; ii, 30°). Similar increases are seen during step displacements of the leg (iii, 5°; iv, 10°). Large displacements (v, IS °) produce an intense phasic discharge that gradually adapts to a lower tonic level. Return movements are followed by an inhibition of background activity paralleling the hysteresis shown by the organ as a whole. (B) Tonic units respond directionally to joint movements (CO). During very rapid repeated displacements (C) tonic activity shows discrete patterning with firing only during movement away from the median range and complete inhibition upon return. Calibration: vertical, 60mV; horizontal, Ai-iv, 0-5s; v, 10a; B, 0-5s; C, 0-2s. Insect proprioception 451 Thus, it appears that some receptors in each group can respond to relaxation as well as stretch of the ligaments, as has been found for the propus-dactyl organ of crustaceans (Mill & Lowe, 1972). For all receptors, however, pulling or releasing the ligaments mimicked their responses to joint movement. Action potential form and the effects ofsoma hyperpolarization. Neither phasic nor tonic units showed large shifts in soma membrane potential during joint movement comparable with generator potentials seen in other receptors, such as the crustacean stretch receptor (Eyzaguirre & Kuffler, 1955). This finding is in agreement with studies of other bipolar neurones such as those of the PD organ of the crab leg (Mendelson, 1963, 1966) and the pit receptors of the crayfish abdomen (Mellon & Kennedy, 1964). In those receptors, however, action potentials often arise abruptly from the baseline, without initial inflections, and are thought to originate in the distal, dendritic process of the cell. In the present study, all stable recordings showed action potentials with distinct initial inflections (Fig. 10). Further experiments are planned to examine the possibility that these pre-potentials represent residues of generator potentials that spread into the soma. Effects of injected current. Phasic and tonic units differed with respect to the effects of depolarizing current injected into the soma. Phasic units responded to current injection with a brief intense discharge that rapidly adapted, even at high levels of current, and abruptly ceased (Fig. 11A). In contrast, tonic units showed sustained increases in discharge rate even to low levels of current (Fig. UBi). Higher levels of injected current produced intense, sustained discharges that were often followed by periods of inhibition of background activity (Fig. UBii). This was particularly pronounced when current was injected over longer periods (Fig. 1 IBiii). Some adap- tation occurred during prolonged constant current injection but sustained increases in tonic activity were followed by periods of inhibition lasting for up to 30 s after cessation of current injection. The effects of depolarizing current injected into the somata of tonic units often matched the effects of step displacements of the tibia. Fig. 12 plots the firingo f a single tonic unit to displacements of two different amplitudes and current injection at two different levels. Large displacements (Fig. 12Ai) and high levels of injected current (Fig. 12Bi) produced increases of activity showing an initial phasic discharge and subsequent sustained elevation of firing rate. Return to the initial position and cessation of current both produced an inhibition of background activity. Small displacements (Fig. 12Aii) and low levels of current injection (Fig. IZBii) were followed by a rapid recovery to baseline levels of activity.

Recordings from axons of chordotonal sensilla To examine more completely the range of responsiveness of individual chordotonal sensilla, intracellular recordings (N=9 animals) were taken from receptor axons close to the metathoracic ganglion. These axons were identified by their intense, short latency firing in response to lift of either the main or flexorligament s of the organ, and consistent responses to subsequent changes in the position of the tibia. It can be noted that all the chordotonal receptor axons were invariably located as a group in a well 452 S. N. ZlLL

Joint angle "V \_

CO J JV L

Fig. 10. Effects of soma hyperpolarization. (A) A phasic unit fires a single action potential to step displacement of the tibia. Progressive injection of hyperpolarizing current (2, 4, 5 nA) produces blocking of the action potential (bridge circuit out of balance to record the response). (B) A tonic unit, held hyperpolarized (3 nA) shows no sustained shift in baseline upon tibial displacement. Calibration: vertical, 20mV; horizontal, A, 50ms, B, 250 ms.

Ai Aii CO- -1 I—

Biii

Fig. 11. Effects of soma depolarization. (A) Phasic units show only transient firing to all levels of depolarizing current. (B) Tonic units show sustained activity to injected current (i,ii). Higher levels of current produce a discharge that adapts slowly over time and is followed by inhibition of background activity (iii). Calibration: vertical, current: A, 3; B, 2nA; voltage: A,B, 40mV; horizontal A.Bi.ii, 0-3 s, Biii, 0-4s. Insect proprioception 453

Displacement

9 18 21 24 27 30

9 18 21 24 27 30 Time (s)

Fig. 12. Comparison of effects of current and displacement. The number of spikes in each successive second are plotted for one tonic unit during displacement and depolarizations at different leveb. (A) Large displacements of the tibia (i) produce a sustained elevation in firing frequency that is followed by inhibition of background activity upon return. Returns after small displacements (ii) show little inhibition. (B) High levels of current injection (i) produce sustained activity that adapts more slowly but is also followed by inhibition of tonic activity. Low levels of current injection (ii) produce only slight rebound inhibition. 454 S. N. Znx defined area of nerve 5 (mid-ventral). This agrees with previous findings of an orderly, somatotopic projection of afferent axons in insect peripheral nerves (Zill, Underwood, Rowley & Moran, 1980). The identification of axons as afferent projections was also confirmed by Lucifer dye injection. Individual phasic and tonic receptors responded only in a defined portion of the range of femoro-tibial joint angle, that is, the organ showed range fractionation (Fig. 13). Phasic units were active in ranges of either flexion or extension but not both. Different phasic units showed considerable variety in the ranges of joint angle within which they responded (maximum, 103°; minimum, 22°), although in general response ranges were quite large. Axons of tonic units proved to be difficult to penetrate individually, presumably due to their small size. All tonic units showed responses in either flexion or extension and increased their discharge as the joint angle was moved away from the rest position, confirming observations from soma recordings. Response ranges of tonic units were smaller than those of phasic units, the minimum being 15° for a tonic unit active only at nearly full flexion. This latter type could be of considerable importance in signalling preparedness for the jump.

PHASIC Respond to i flexion jj iii Respond to i extension jj

0 20 40 60 80 100 120 140 160

20-, B TONIC

~» 1 .OH \ \ I > I \ f A / 20 40 60 80 100 120 140 160 Joint angle (degrees)

Fig. 13. Range fractionation. The response ranges of five phasic and three tonic units recorded intracellularly from a single animal are plotted on the abscissa (0° = full flexion, 170° = full extension). Phasic units had large response ranges. Tonic units had smaller response ranges and increased their responsefrequenc y (plotted on the ordinate) in the extreme ranges of joint angle. Insect proprioception 455

DISCUSSION Responses of chordotonal sensilla The present study has shown that the femoral chordotonal organ contains four types of receptors, all directionally sensitive: tonic receptors, responding in ranges of either joint flexion or extension, and phasic receptors that fire only in response to flexion or extension movements. These basic types of sensilla have been found in every chordotonal organ that has been studied physiologically (Wiersma & Boettiger, 1959; Bush, 1965). Individual receptors in joint chordotonal organs thus invariably indicate direction as well as joint movement. Certain phasic receptors responded to all velocities of joint movements with only single action potentials. These units are similar to phasic receptors found in other chordotonal organs (Wiersma & Boettiger, 1959) that apparently function as extremely sensitive movement detectors. Other phasic receptors showed an increase in firing frequency that was linearly related to the velocity of joint movement but showed saturation during very rapid movements. Similar saturation of responsiveness has been found in units of the propus-dactyl (PD) chordotonal organ of the crab leg (Mill & Lowe, 1972) that may show a maximum response of only 24-25 Hz at velocities as low as 5 mm s~'. Other chordotonal organs of the crab leg show saturation at higher response frequencies (Bush, 1965) similar to those found for the locust femoral chordotonal organ in the present study. It can be noted that, in the locust, limits to response frequency are not set by refractoriness of the spike-transmitting capacities of the soma as direct depolarization of the cell bodies of chordotonal sensilla produced much higher firing frequencies than did mechanical stimulation. Rather, as postulated by Wiersma (in Mendelson, 1963), these limitations in frequency are probably due to properties of the mechano-electric transducer membrane of the receptor dendrite. All tonic units of the femoral chordotonal organ progressively increased their firing rate in positions away from the median range of joint angle and reached maximum activity in the extremes of tibial position. Similar types of responses have been found in other arthropod chordotonal organs (Cohen, 1963; Bush, 1965). This type of response provides the maximum changes in frequency in ranges farthest from the resting position of the joint. Tonic units of chordotonal organs of other arthropods can, however, show more complex changes in firing frequency than those found in the locust (U. Bassler, personal communication). Tonic units of the femoral chordotonal organ also showed some phasic discharges to joint movement, similar to the 'intermediate' type units found in organs of crab legs (Wiersma & Boettiger, 1959; Clarac, 1968), but those organs also contain some pure tonic units that do not possess phasic properties, a type not found in the locust. It should be noted that many tonic receptors had quite small somata so it is possible that such units are present in the organ but remained undetected. If so, they probably form only a small percentage of the total receptors of the organ. 456 S. N. ZILL In sum, the characteristics of the sensilla of the locust femoral chordotonal organ found in the present study are clearly within the spectrum of types shown by other arthropod chordotonal organs. The specific responses of the receptors in the locust, however, may reflect the specialized functions of the organ (see below).

Grouping of receptors The sensilla of the metathoracic femoral chordotonal organ were found to be consistently organized into two groups, based upon their dendritic and scolopale orientations and their directional sensitivity. Spatial separation according to direction of response has been shown for other leg chordotonal organs. For example, in the PD organ of the crab leg flexion-an d extension-sensitive cells are found on opposite sides of the receptor strand (Wiersma & Boettiger, 1959; Hartman & Boettiger, 1967). Anatomical grouping of receptors according to dendritic orientation has been found in other insect leg chordotonal organs (Young, 1970) and is a prominent feature of sensilla associated with auditory receptors (Doolan & Young, 1981). In those organs, it has been postulated that anatomical orientation determines the specific range of responsiveness, although this has not yet been specifically examined.

Adequate stimuli for chordotonal sensilla The present study has shown that each of the two ligaments of the metathoracic femoral chordotonal organ selectively produce tonic responses in different ranges of joint angle: the main ligament produces tonic afferent discharges in ranges of joint flexion (0-80°) while the flexor ligament mediates responses in ranges of joint extension. As the receptors of the organ also show a consistent separation according to their directional sensitivity and range of responsiveness it would appear that each of these ligaments selectively acts upon different groups of sensilla: stretch of the main ligament produces tonic activity in the dorsal group of sensilla, stretch of the flexor ligament elicits responses from tonic units in the ventral group. These conclusions are supported by those experiments in which ligaments were directly stretched, producing selective activation of tonic receptors. Further, as each of these groups of receptors shows a consistent dendritic and scolopale orientation, the adequate mechanical stimuli for the tonic receptors would appear to be as follows: the dorsal group of receptors are clearly orientated so as to be stretched by pulling of the main ligament; the ventral group of receptors, whose dendrites terminate close to the flexor ligament, should be both stretched and bent by this attachment. Thus both stretching and bending of the dendrite would seem to be adequate stimuli for tonic receptors. The effects of the chordotonal ligaments upon phasic receptors may be more complex. Phasic units of the dorsal group that respond in ranges of joint flexion and those receptors of the ventral group active in ranges of extension, should respond to stretching and bending, respectively, as do the tonic units of those groups. However, the dorsal group of receptors also contains phasic units responding to flexion movements at angles greater than 80° while the ventral group possesses receptors active at joint angles less than 80°. These receptors may respond to ligament Insect proprioception 457 relaxation rather than stretch as shown by those experiments in which responses were recorded while the insertions of the organ were directly manipulated. Relaxation- sensitive receptors have been conclusively identified in the PD organ of the crab (Mill & Lowe, 1972, 1973), and have been shown to possess morphologically specialized dendritic terminations. However, in the locust there may also be some interaction of stretch of each of the ligaments to affect a small number of receptors in both groups. For example, the flexor ligament may be responsible for discharges elicited in ranges of extreme extension from receptors close to the attachment of the organ. Also, experiments examining phasic responses to joint displacements when individual ligaments had been severed produced equivocal results due to unavoidable slight movements of the cut ligaments. For the majority of receptors of the organ, however, individual pulls to the main or flexor ligaments effectively mimicked the effects of joint displacements. Unfortunately, there is no clear agreement in the literature about the nature of sensory transduction in chordotonal organs to permit a simple analysis of the mechanical forces that affect these receptors. Wiersma & Boettiger (1959) first proposed that stretch of the dendrites was the effective stimulus for receptors of the crab PD organ, a view that has subsequently been adopted for the myochordotonal organ (Cohen, 1963) and for crustacean antennal organs (Wyse & Maynard, 1965). In contrast, Mendelson (1963), Bush (1965) and Taylor (1967) have suggested that flexion or bending of the dendrite is the adequate stimulus. Further, as noted above, individual units may respond to relaxation as well as stretch (Mill & Lowe, 1972). The results of the present study suggest that each of these views may be correct and that, depending upon the mechanical arrangement of sensilla, individual receptors may respond to stretching, bending or relaxation. Direct observation of dye-filled receptors might aid in understanding the anatomical basis for differential responses in receptors of the femoral chordotonal organ.

Interpretation of behavioural experiments The results of the present study clarify the specific effects of the behavioural experiments of Bassler (1968, 1979) in which the ligaments of the organ were ablated or their insertions moved. First, Bassler (1968) showed that jumping could not be elicited after cutting the main ligament of the organ. The present study has shown that the effect of this operation is to eliminate responses in ranges of joint flexion. The discharge of the organ, however, does not indicate full tibial extension when the joint is flexed, as assumed by Heitler & Burrows (1977), but remains equivalent to that seen at a joint angle of 80°. Ablation of the main ligament also eliminates the specific input of the chordotonal organ during the co-contraction phase prior to the jump. This finding supports the view (Pearson et al. 1980) that the femoral chordotonal organ provides a substantial contribution to triggering of the jump. Bassler (1979) also showed that switching the insertion of the main ligament of the chordotonal organ, so that it was stretched by joint extension rather than flexion, has discrete effects upon posture and locomotion. After this operation, animals would 458 S. N. ZILL often hold the tibia in full extension, even while walking. The present study suggests that the effect of such alteration of the main ligament would be to reverse the input from tonic units in the dorsal group of receptors, exciting these sensilla upon joint extension rather than joint flexion. Any slight extension movements would then be perceived as joint flexions by the animal. The reflex compensatory response to perceived flexion would be joint extension that would then further excite the dorsal receptors. The compensatory system would now be caught in continuous positive feedback which would generate complete joint extension. It should be noted that the responses of the ventral group of receptors should remain unaltered. The animal is, thus, apparently incapable of overcoming erroneous input from one group of receptors with that provided by another. The next paper in this series also shows that potent reflex effects can be elicited by excitation of only one of the groups of receptors of the chordotonal organ (Zill, 1985).

Cellular properties of individual sensilla One finding of the present study differs from previous accounts of responses of bipolar neurones (Mendelson, 1963, 1966; Mellon & Kennedy, 1964). Moderate levels of depolarizing current injection into the somata of tonic units produced a sustained increase in firing rate that was followed by an inhibition of background activity. While the effects of soma depolarization were not extensively examined, Mellon & Kennedy found only phasic discharges in response to extracellular stimulation of the pit receptors, although it was not determined whether these units respond tonically to mechanical stimuli or phasically to small vibrations (Mellon, 1963). Mendelson (1963, 1966) used current injection in studies of the PD receptors of the crab leg, but does not mention any post-excitatory inhibition, although he did not systematically classify units as phasic or tonic. These differences are explicable if it is assumed that current passed into the soma of tonic units of the locust chordotonal organ spreads and directly affects the distal spike initiating zone. It should-again be noted that the dendrites of tonic units were often quite short (often less than 75 Jim). In crustaceans, dendrites of bipolar receptors are quite long, commonly exceeding 200 /im. In those receptors, high levels of injected current would be required to affect the distal dendrite. The proximity of the dendritic end to the soma in locust chordotonal sensilla may prove useful for studying the mechanisms underlying transduction in these receptors.

Hysteresis The tonic units of the femoral chordotonal organ showed pronounced hysteresis: leg movements towards the median range of joint angle were accompanied by a substantial inhibition of activity followed only slowly by recovery to normal background firing. Other insect receptors show similar directional hysteresis, such as the mesothoracic femoral chordotonal organ (Burns, 1974) and the multipolar receptors of the metathoracic joint (Coillot & Boistel, 1969). The cellular mechanisms underlying this phenomenon are unknown. Burns (1974) speculated that in the Insect proprioception 459 mesothoracic organ, the mechanical link to the flexortibi a muscle produced hysteresis due to viscosity in the muscle fibres. This is apparently not the case for the metathoracic organ as there is no direct ligament to flexor muscle fibres. Rather, the results of soma depolarization suggest that refractoriness of the spike-initiating mechanism may contribute to hysteresis. Intense firing produced by either current injection or large joint movements was followed by periods of inhibition of background activity upon return. Unfortunately, it was not possible to study the effects of very large joint movements so that a strict comparison between firing rate and subsequent inhibition for both movement and current injection could not be obtained. It should be noted that other mechanisms may also contribute to hysteresis, such as the time course of mechanical recovery of dendritic deformation or the properties of the mechano-transducer membrane itself.

Potential functions ofsensilla of the femoral chordotonal organ The characteristics of receptors of the locust femoral chordotonal organ may be considered adaptations for the specialized uses and properties of the metathoracic leg and suggest the following functions for this receptor. (1) Detection of joint movement. All the sensilla of the chordotonal organ are highly sensitive to joint movement. Some phasic receptors, with large somata and axons, apparently only detect joint movement but these can rapidly inform the nervous system of changes in joint angle. As succeeding studies in this series will show, this sensitivity permits the locust rapidly to respond to imposed movements either by compensation or by withdrawal of the leg when compensation is undesirable. (2) Detection of readiness for the jump. One clear specialization of the organ, that is a consequence of the anatomy of the femoro-tibial joint, is its response to tibial bending produced by contractions of the extensor muscle when the leg is fully flexed. Further, because of the fractionation of response range, some individual sensilla may selectively provide the nervous system with precise information as to whether sufficient energy is stored in deformation of the cuticle to execute successfully a jump or defensive kick. (3) Assisting in load compensation. As will be shown in the next study in this series, one of the key functions of the chordotonal organ is to produce reflex discharges in leg motoneurones to compensate for changes in load. There are several characteristics of individual receptors of the organ that precisely attune it to this function. First, all tonic and most phasic units show discharges that are sensitive to the velocity of joint movement. Purely tonic receptors, that would respond only to joint angle, were not found in the organ. By increasing frequency to increasing velocity of movement, the organ provides an input to motoneurones that signals the appropriate level of activity needed for compensation. Further, all tonic units increase their discharge rate in ranges furthest from the median joint angle. These units also show the greatest hysteresis at these extremes. It can be noted that these characteristics provide the greatest change in afferent frequency in joint positions that require the largest changes in motoneurone firing rate for compensation due to the declining mechanical advantage of the tibial muscles (Burrows & Horridge, 1974). 460 S. N. ZILL In sum, the properties of individual sensilla of the femoral chordotonal organ suggest that these receptors are specialized to fulfill discrete functions in the locust's behaviour. Many limb proprioceptive sense organs may also be particularly adapted to match the anatomical and physiological characteristics of the joints and muscles they innervate.

This work was supported by NSF Grant SPI-7914916 and NIH Grant 5F32NS06373. I thank Graham Hoyle and Eric Schabtach for helpful comments on the manuscript and Suzanne Royer for assisting with histological preparations.

REFERENCES BASSLER, U. (1968). ZursteuerungdesspriagensbeiderWanderheuschrecke.ScAMtocerca^i'^ana.A^ftemeftA 4, 112. BASSLER, U. (1979). Effects of crossing the receptor apodeme of the femoral chordotonal organ on walking, jumping and singing in locusts and grasshoppers. J. comp. Physiol. 134, 173-176. BAXENDALE, R. H. & FERRELL, W. R. (1981). The effect of knee joint afferent discharge on transmission in flexor reflex pathways in decerbrate cats. J. Physiol., Land. 315, 231-242. BENNET-CLAJUC, H. C. (1975). The energetics of the jump of the locust, Schistocercagregaria. J'. exp. Biol. 63, 53-83. BOYD, I. A. & ROBERTS, T. D. M. (1953). Proprioceptive discharges from stretch receptors in the knee joint of the cat. J. Physiol. Land. 122, 38-58. BURKE, W. (1954). An organ for proprioception and vibration sense in Cardnus maenus. J. exp. Biol. 31, 127-138. BURNS, M. D. (1974). Structure and physiology of the locust femoral chordotonal organ. J. Insect Physiol. 20, 1319-1339. BURROWS, M. & HoRRIDGE, G. A. (1974). The organization of inputs to motoneurons of the locust metathoracic leg. Phil. Trans. R. Soc. Ser. B 269, 49-94. BUSH, B. M. H. (1965). Proprioception by the coxo-basal chordotonal organ, CB, in legs of the crab, Cardnus maenas.J. exp. Biol. 42, 285-297. CAMPBELL, J. I. (1961). The anatomy of the nervous system of the mesothorax of Locusta migratoria migrator/rides R. and F. Proc. zool. Soc., Land. 137, 403-432. CLARAC, F. (1968). Proprioception by the ischio-meropodite region in the legB of the crab, Cardnus mediterrancus. (C.) Z. vergl. Physiol. 61, 224-245. COHEN, M. J. (1963). The crustacean myochordotonal organ as a proprioceptive system. Comp. Biochem. Physiol. 8, 223-243. ConxOT, J. P. & BoiSTEL, J. (1969). Etude de l'activite propagee de recepteurs a retirement de la patte methathoracique du criquet, Schistocerca gregaria. J. Insect Physiol. IS, 1449-1470. DEBAlsmux, P. (1938). Organes scolopidiaux des pattes des insects. II. Cellule 47, 77-202. DOOLAN, J. M. & YOUNG, D. (1981). The organization of the auditory organ of the bladder cicada, Cystosoma saundersii. Phil. Trans. R. Soc. Ser. B 291, 525-540. EYZAOUIRRE, C. & KUPFLER, S. W. (1955). Process of excitation in the dendrites and in the soma of single isolated sensory nerve cells of the lobster and crayfish. J. gen. Physiol. 39, 87-119. FERRELL, W. R. (1977). The discharge of mechanoreceptors in the cat knee joint at intermediate angles. J. Physiol., bond. 268, 23-24P. GRAHAM, D. & BASSLER, U. (1981). Effects of afference sign reversal on motor activity in walking stick insects (Carausius morosus).J. exp. Biol. 91, 179-183. GRANIT, R. (1955). Receptors and Sensory Perception. New Haven, Connecticut: Yale University Press. GRIOG,P.,HARRIGAN, E. P. &FOOARTY, K. E. (1978). Segmental reflexes mediated by joint afferent neurons in cat knee. J. Neurvphysiol. 41, 9-14. HARTMAN, H. B. & BoETTlGER, E. G. (1967). The functional organization of the propus-dactylus organ in Cancer irmratus Say. Comp. Biochem. Physiol. 22, 651-662. HEITLER, W. J. (1977). The locust jump. III. Structural specializations of the metathoracic tibiae. J. exp. Biol. 67, 29-36. HEITLER, W. J. & BURROWS, M. (1977). The locust jump. II. Neural circuits of the motor programme. J. exp. Biol. 66, 221-241. Insect proprioception 461

HOYLE, G. & BUKKOWS, M. (1973). Neural mechanisms underlying behavior in the locust, ScMstocerca gregaria. I. Physiology of identified motorneurons in the metathoracic ganglion..7. Neurobiol. 4, 3—41. HUSTEKT, R. (1982). The proprioceptive function of a complex chordotonal organ associated with the mesothoracic coxa in locuets. J. comp. Physiol. 147, 389-399. KENNEDY, D. (1966). The comparative physiology of invertebrate central neurons. Adv. comp. Physiol. Biochem.Z, 117-184. LINDSEY, B. G. & BROWN, H. K. (1982). Convergence of parallel sensory channels on crayfish claw motor neurons. Changing firing probabilities and synaptic effects of simultaneously monitored proprioceptors. J. Neurophysiol. 47, 1144-1159. LOWE, D. A. & MILL, P. J. (1972). The relationship between the PD proprioceptor, the propodite-dactylopodite joint and the dactylopodite flexor muscle in the walking legs of Cancer pagurus. Mar. Behavl Physiol. 1, 157-170. MATTHEWS, P. B. C. (1972). Mammalian Muscle Receptors andTheirCentral Actions. London: Edward Arnold Ltd. MELLON, D. (1963). Electrical responses from dually innervated tactile receptors on the thorax of the crayfish. J. exp. Biol. W, 137-148. MELLON, D. & KENNEDY, D. (1964). Impulse origin and propagation in a bipolar sensory neuron. J. gen. Physiol. 47, 487-499. MENDELSON, M. (1963). Some factors in the activation of crab movement receptors. J. exp. Biol. 40, 157-169. MENDELSON, M. (1966). The site of impulse initiation in bipolar receptor neurons of Callinectes sapidus L. jf.exp. Biol. 45, 411-420. MILL, P. J. (1976). Chordotonal organs of crustacean appendages. In Structure and Function of Proprioceptors in the Invertebrates, (ed. P. J. Mill), pp. 243-297. London: Chapman & Hall. MILL, P. J. & LOWE, D. A. (1972). An analysis of the types of sensory unit present in the PD proprioceptor of decapod crustaceans. J. exp. Biol. 56, 509-525. MILL, P. J. & LOWE, D. A. (1973). The fine structure of the PD proprioceptor of Cancer pagurus. I. The receptor strand and the movement sensitive cells. Proc. R. Soc. Ser. B 184, 179-197. PABST, H. & KENNEDY, D. (1967). Cutaneous mechanoreceptors influencing motor output in the crayfish abdomen. Z. vergl. Physiol. 57, 190-208. PEARSON, K. G., HEITLER, W. J. & STEEVES, J. D. (1980). Triggering of the locust jump by multimodal inhibitory interneurons. J. Neuwphystol. 43, 257-278. SANDEMAN, D. C. (1968). A sensitive position measuring device for biological systems. Comp. Biochem. Physiol. 24, 635-638. SCHNORBUS, H. (1971). Die subgenualen Sinnesorgane von Periplaneta americana: Histologie und Vibra- tionsschwellen. Z. vergl. Physiol. 71, 14-48. STEEVES, J. D. & PEARSON, K. G. (1982). Proprioceptive gating of inhibitory pathways to hindleg flexor motoneurons in the locust. J. comp. Physiol. 146, 507-515. TAYLOR, R. C. (1967). The anatomy and adequate stimulation of a chordotonal organ in the antennae of a hermit crab. Comp. Biochem. Physiol. 20, 709-717. USHERWOOD, P. N. R., RUNION, H. I. & CAMPBELL, J. I. (1968). Structure and physiology of a chordotonal organ in the locust leg. J. exp. Biol. 48, 305-323. WlERSMA, C. A. G. & BoETTIGER, E. G. (1959). Unidirectional movement fibres from a proprioceptive organ of the crab, Cardnus maenas.J. exp. Biol. 36, 102—112. WIERSMA, C. A. G., VAN DER MARK, F. &FIORE, L. (1970). On the firing pattern of the "movement" receptors of the elastic organ of the crab, Cardnus. Comp. Biochem. Physiol. 34, 833—840. WILSON, D. M. (1961). The central nervous control of flight in a locust.J exp. Biol. 38, 471-490. WYSE, C. A. & MAYNARD, D. M. (1965). Joint receptors in the antennule otPanulirus argus Latreille.7- exp. Biol. 42, 521-535. YOUNG, D. (1970). The structure and function of a connective chordotonal organ of the cockroach leg. Phil. Trans. R. Soc. Ser. B 256, 401-428. YOUNG, D. & BALL, E. (1973). Structure and development of the auditory system in the prothoracic leg of the cricket, Teleogryllus commodus (Walker). 1. Adult structure. Z. Zellforsch. mikrosk. Anat. 147, 293-312. ZILL, S. (1985). Plasticity and proprioception in insects. II. Modes of reflex action of the locust metathoracic femoral chordotonal organ, jf. exp. Biol. 116, 463-480. ZILL, S. N., UNDERWOOD, M. A., ROWLEY, J. C. & MORAN, D. T. (1980). A somatotopic organization of afferents in insect peripheral nerves. Brain. Res. 198, 253-269.