Segmentally distributed metamorphic changes in neural circuits controlling abdominal bending in the hawkmoth Manduca sexta

During the metamorphosis of Manduca sexta the larval nervous system is reorganized to allow the generation of behaviors that are specific to the pupal and adult stages. In some instances, metamorphic changes in neurons that persist from the larval stage are segment-specific and lead to expression of segment-specific behavior in later stages. At the larval-pupal transition, the larval abdominal bending behavior, which is distributed throughout the abdomen, changes to the pupal gin trap behavior which is restricted to three abdominal segments. This study suggests that the neural circuit that underlies larval bending undergoes segment specific modifications to produce the segmentally restricted gin trap behavior. We show, however, that non-gin trap segments go through a developmental change similar to that seen in gin trap segments. Pupal-specific motor patterns are produced by stimulation of sensory neurons in abdominal segments that do not have gin traps and cannot produce the gin trap behavior. In particular, sensory stimulation in non-gin trap pupal segments evokes a motor response that is faster than the larval response and that displays the triphasic contralateral-ipsilateral-contralateral activity pattern that is typical of the pupal gin trap behavior. Despite the alteration of reflex activity in all segments, developmental changes in sensory neuron morphology are restricted to those segments that form gin traps. In non-gin trap segments, persistent sensory neurons do not expand their terminal arbors, as do sensory neurons in gin trap segments, yet are capable of eliciting gin trap-like motor responses. Accepted: 10 January 1997     

Neural correlates of habituation of the proleg withdrawal reflex in larvae of the hawk moth, Manduca sexta

The larval proleg withdrawal reflex of the hawk moth, Manduca sexta, exhibits robust habituation. This reflex is evoked by deflecting one or more mechanosensory planta hairs on a proleg tip. We examined neural correlates of habituation in an isolated proleg preparation consisting of one proleg and its segmental ganglion. Repeated deflection of a single planta hair caused a significant decrease in the number of action potentials evoked in the proleg motor nerve (which carries the axons of proleg retractor motor neurons). Significant response decrement was seen for interstimulus intervals of 10 s, 60 s and 5 min. Response decrement failed to occur in the absence of repetitive stimulation, the decremented response recovered spontaneously following a rest, and electrical stimulation of a body wall nerve facilitated the decremented response (a neural correlate of dishabituation). Adaptation of sensory neuron responses occurred during repeated hair deflections. However, when adaptation was eliminated by direct electrical stimulation of sensory neurons, the response in the proleg motor nerve still decreased significantly. Muscle recordings indicated that the response of an identified proleg retractor motor neuron decreased significantly during habituation training. Thus, habituation of the proleg withdrawal reflex includes a central component that is apparent at the level of a single motor neuron. Accepted: 20 December 1996     

Development of the gin trap reflex inManduca sexta: a comparison of larval and pupal motor responses

1. Responses of motor neurons in larvae and pupae of Manduca sexta to stimulation of tactile sensory neurons were measured in both semi-intact, and isolated nerve cord preparations. These motor neurons innervate abdominal intersegmental muscles which are involved in the production of a general flexion reflex in the larva, and the closure reflex of the pupal gin traps. 2. Larval motor neurons respond to stimulation of sensory neurons innervating abdominal mechanosensory hairs with prolonged, tonic excitation ipsilaterally, and either weak excitation or inhibition contralaterally (Figs. 4A, 6). 3. Pupae respond to tactile stimulation of mechanosensory hairs within the gin traps with a rapid closure reflex. Motor neurons which innervate muscles ipsilateral to the stimulus exhibit a large depolarization, high frequency firing, and abrupt termination (Figs. 2, 4B). Generally, contralateral motor neurons fire antiphasically to the ipsilateral motor neurons, producing a characteristic triphasic firing pattern (Figs. 7, 8) which is not seen in the larva. 4. Pupal motor neurons can also respond to sensory stimulation with other types of patterns, including rotational responses (Fig. 3A), gin trap opening reflexes (Fig. 3B), and 'flip-flop' responses (Fig. 9). 5. Pupal motor neurons, like larval motor neurons, do not show oscillatory responses to tonic current injection, nor do motor neurons of either stage appear to interact synaptically with one another. Most pupal motor neurons also exhibit i-V properties similar to those of larval motor neurons (Table 1; Fig. 10). Some pupal motor neurons, however, show a marked non-linear response to depolarizing current injection (Fig. 11).     

Physiological characterization of antennular flicking reflexes in the crayfish

The cellular substrates of antennular flicking behavior in the crayfish Procambarus clarkii were investigated. Flicking involves fast downward movements of the external filament of each biramous antennule (1st antenna), and is mediated by phasic contractions of a short muscle, the external filament depressor. Phasic contractions of the external filament depressor depend upon stereotyped impulse bursts in a single motorneuron (P1). These bursts have a characteristic impulse frequency profile that is consistent upon successive occurrences. The temporal characteristics of the impulse burst suggest that the central depolarizations generating each burst may be similar to driver potentials described for motor neurons in crustacean cardiac ganglia. Responses of the external filament to odorants have a long latency and are characterized by repetitive bursts and tonic activity in some external filament depressor fibers. Tonic activity in a slowly contracting muscle, the antennular depressor muscle, is also evoked by chemical stimulation. Flicking is consistently evoked only by mechanical or hydrodynamic stimulation of the cephalothorax, antennae and antennules. The sensitivity and short latency of the hydrodynamic antennule-generated flick reflex is consistent with the sensitivity of rapidly conducting, hydrodynamically activated mechanoreceptor neurons in both antennular filaments. I propose that antennular flicking, which has been shown to enhance the dynamic response characteristics of olfactory receptor neurons on the external antennular filament, has evolved as a response to the turbulence associated with fluid movement, within which chaotic odorant concentration fronts may be imbedded. Accepted: 23 October 1996     

Morphology and physiology of a hair plate sensory organ located on the antenna of the rock lobster Palinurus vulgaris

At the level of the J1 joint of each antenna of the rock lobster Palinurus vulgaris a hair plate sensory organ (hp) similar to those described in insects has been observed. The hp is located on the internal side of the S2 segment of the antenna, close to the soft articulating membrane of the J1 joint. It is formed by a triangular surface of cuticle about 3mm2 in area, covered with numerous hairs of different lengths (Figs. 1 and 2). Details of the hp were studied by scanning electron microscopy (Fig. 2). Physiological stimulation of the hp hairs occurs during medial movement of the J1 joint. Under this condition the soft articulating membrane rolls over the hairs and bends them progressively back onto the cuticle. Flexion of all the hairs corresponds to a medial movement of the J1 through 40 degrees. During this type of movement, the number of successively flexed hairs increases linearly (Fig. 3). Electrophysiological recordings of the hp sensory nerve correlated with selective mechanical stimulation of individual hairs demonstrated that each hair is innervated by a single sensory fiber. This sensory neurone responds phasically when the hair is flexed back onto the cuticle (as during an S2 medial movement) and when it returns to its resting position (as during an S2 lateral movement). Most of the sensory neurones are sensitive to the movement velocity of the hairs (Figs. 4 and 5). When the hair is maintained flexed its sensory neurone discharges tonically (Fig. 4). Electrical stimulation of the hp sensory nerve induced reflex actions in the external and internal rotator muscles of the segment S1. These effects were found to selectively activate the tonic motor command of these muscles (Fig. 6).     

Non-spiking local interneurones mediate abdominal extension related descending control of uropod motor neurones in the crayfish terminal abdominal ganglion

The role of non-spiking local interneurones in the synaptic interactions between abdominal extension-evoking descending interneurones and uropod motor neurones in the terminal abdominal ganglion of the crayfish Procambarus clarkii (Girard) was investigated electrophysiologically. Continuous electrical stimulation of the lateral region of the 3rd-4th abdominal connective that included abdominal extension evoking interneurones excited the opener motor neurones and inhibited the closer, reductor motor neurone. Spikes from a single descending interneurone evoked consistent and short latency (0.8–0.9 ms) excitatory postsynaptic potentials (e.p.s.ps) in the opener motor neurones, and evoked rather long-latency (1.5–2.7 ms) inhibitory postsynaptic potentials (i.p.s.ps) in the reductor motor neurone. Many non-spiking interneurones also received depolarizing p.s.ps (0.8–2.5 ms in latency) that were usually faster than i.p.s.ps of the reductor motor neurone if both neurones were recorded sequentially in the same preparation. Non-spiking interneurones received convergent inputs from several descending interneurones and made inverting connection with the reductor motor neurone. Elimination of descending inputs to a particular non-spiking interneurone could reduce the inhibitory response of the reductor motor neurone. These observations strongly suggested that descending inhibitory inputs to the closer, reductor motor neurone were mediated by non-spiking interneurones. Furthermore, some non-spiking interneurones made output connections with the opener motor neurones. The disynaptic pathway through non-spiking interneurones is significant to control and modulate the opening pattern of the uropod during abdominal extension. Accepted: 27 December 1996     

Spinal reflexes in the long-tailed stingray, Himantura fai

We have exploited the segregation of motor and sensory axons into peripheral nerve sub-compartments to examine spinal reflex interactions in anaesthetized stingrays. Single, supra-maximal electrical stimuli delivered to segmental sensory nerves elicited compound action potentials in the motor nerves of the stimulated segment and in rostral and caudal segmental motor nerves. Compound action potentials elicited in segmental motor nerves by single stimuli delivered to sensory nerves were increased severalfold by prior stimulation of adjacent sensory nerves. This facilitation of the segmental reflex produced by intense conditioning stimuli decreased as it was applied to more remote segments, to approximately the same degree in up to seven segments in the rostral and caudal direction. In contrast, an asymmetric response was revealed when test and conditioning stimuli were delivered to different nerves, neither of which was of the same segment as the recorded motor nerve: in this configuration, conditioning volleys generally inhibited the responses of motoneurons to stimuli delivered to more caudally located sensory nerves. This suggests that circuitry subserving trans-segmental interactions between spinal afferents is present in stingrays and that interneuronal connections attenuate the influence that subsequent activity in caudal primary afferents can have on the motor elements.     

Dynamic responses of series force receptors innervating the opener muscle apodeme in the blue crab, Callinectes sapidus

Receptors monitoring muscle force innervate the opener muscle apodeme in the walking legs of the blue crab, Callinectes sapidus. Biocytin backfills reveal 9–15 bipolar neurons with somata as large as 60 μm positioned at the distal end of the apodeme. Sensory endings insert into the apodeme and are in series with the opener muscle. The axons of these neurons form the opener apodeme sensory nerve that merges with the most distal branch of the opener motor nerve. Recordings reveal that the receptors are not spontaneously active nor do they respond to passive muscle stretch. Isometric muscle contraction evoked by stimulating the opener excitor motor neuron is the adequate stimulus for receptor firing. Most significant is the finding that during contraction, over a wide range of forces, the firing rate of individual receptors closely parallels the rate of change of isometric force. The peak instantaneous frequency typically occurs at the force derivative maximum, but not at maximum force development. Thus, receptors of the opener apodeme sensory nerve more closely monitor changes in isometric force rather than the total force achieved. Accepted: 20 September 1996     

The tarso-pretarsal chordotonal organ as an element in cockroach walking

Many types of sense organs have been demonstrated to show repetitive discharges during walking that could provide informational cues about leg movements and other parameters of locomotion. We have recorded activities of receptors of the distal (tarsal) segments of the cockroach hindleg in restrained and freely moving animals while they were videotaped. These recordings show peaks of activities at the onset and termination of the stance phase. We have morphologically and physiologically identified a joint angle receptor, the tarso-pretarsal chordotonal organ, that contributes to the discharges seen late in stance, prior to the onset of leg flexion in swing. This sense organ encodes the angle and rate of change of the most distal leg joint and specifically discharges when the claws are disengaged from the substrate. Applied displacements of the claws in restrained preparations elicit reflex activation of the tibial flexor muscle and a crossed extensor reflex in the opposite hindleg. These reflexes could function to insure that leg flexion in swing does not occur until the claws are disengaged and to enhance support by the opposite hindleg. Thus, the regular discharges of the chordotonal organ could assure efficient and coordinated muscle contractions and movements during normal, unperturbed walking. Accepted: 2 January 1997     

The pattern of sensory discharge can determine the motor response in young Xenopus tadpoles

Young Xenopus tadpoles were used to test whether the pattern of discharge in specific sensory neurons can determine the motor response of a whole animal. Young Xenopus tadpoles show two main rhythmic behaviours: swimming and struggling. Touch-sensitive skin sensory neurons in the spinal cord of immobilised tadpoles were penetrated singly or in pairs using microelectrodes to allow precise control of their firing patterns. A single impulse in one Rohon-Beard neuron (= light touch) could sometimes trigger “fictive” swimming. Two to six impulses at 30–50 Hz (= a light stroke) reliably triggered fictive swimming. Neither stimulus evoked fictive struggling. Twenty-five or more impulses at 30–50 Hz (= pressure) could evoke a pattern of rhythmic bursts, distinct from swimming and suitable to drive slower, stronger movements. This pattern showed some or all the characteristics of “fictive” struggling. These results demonstrate clearly that sensory neurons can determine the pattern of motor output simply by their pattern of discharge. This provides a simple form of behavioural selection according to stimulus. Accepted: 28 November 1996     

Intersegmental interneurons serving larval and pupal mechanosensory reflexes in the moth Manduca sexta

1. Intersegmental interneurons (INs) that participate in the larval bending reflex and the pupal gin trap closure reflex were identified in the isolated ventral nerve cord of Manduca sexta. INs 305, 504, and 703 show qualitatively different responses in the pupa than in the larva to electrical stimulation of sensory neurons that are retained during the larval-pupal transition to serve both reflexes. Action potentials produced by current injected into the 3 interneurons excite motor neurons that are directly involved in the larval and pupal reflexes. The excitation of the motor neurons is not associated with EPSPs at a fixed latency following action potentials in the interneurons, and thus there do not seem to be direct synaptic connections between the interneurons and the motor neurons. 2. IN 305 (Fig. 2) has a lateral soma, processes in most of the dorsal neuropil ipsilateral to the soma, and a crossing neurite that gives rise to a single contralateral descending axon. IN 305 is excited by stimulation of the sensory nerve ipsilateral to its soma in the larva and the pupa. Stimulation of the sensory nerve contralateral to its soma produces an inhibitory response in the larva, but a mixed excitatory/inhibitory response to the identical stimulus in the pupa. 3. IN 504 (Fig. 3) has a lateral soma, processes throughout most of the neuropil ipsilateral to the soma, and a crossing neurite that bifurcates to give rise to a process extending to the caudal limit of the neuropil and an ascending axon. IN 504 is excited by stimulation of the sensory nerve ipsilateral to its soma in both larvae and pupae, while the response to stimulation of the sensory nerve contralateral to its soma is inhibitory in the larva but mixed (excitatory/inhibitory) in the pupa. 4. IN 703 has a large antero-lateral soma, a neurite that extends across to the contralateral side giving rise to processes located primarily dorsally in both ipsilateral and contralateral neuropils, and two axons that ascend and descend in the connectives contralateral to the soma (Fig. 4). IN 703 responds to stimulation of the sensory nerves on either side of the ganglion, but the form of the response changes during the larval-pupal transition. In the larva, the response consists of very phasic (0-2 spikes) excitation, but in the pupa there is a prolonged excitation that greatly outlasts the stimulus (Fig. 6).(ABSTRACT TRUNCATED AT 400 WORDS)     

Directionality of auditory nerve fiber responses to pure tone stimuli in the grassfrog, Rana temporaria. II. Spike timing

We studied the directionality of spike timing in the responses of single auditory nerve fibers of the grass frog, Rana temporaria, to tone burst stimulation. Both the latency of the first spike after stimulus onset and the preferred firing phase during the stimulus were studied. In addition, the directionality of the phase of eardrum vibrations was measured. The response latency showed systematic and statistically significant changes with sound direction at both low and high frequencies. The latency changes were correlated with response strength (spike rate) changes and were probably the result of directional changes in effective stimulus intensity. Systematic changes in the preferred firing phase were seen in all fibers that showed phaselocking (i.e., at frequencies below 500–700 Hz). The mean phase lead for stimulation from the contralateral side was approximately 140° at 200 Hz and decreased to approximately 100° at 700 Hz. These phaseshifts correspond to differences in spike timing of approximately 2 ms and 0.4 ms respectively. The phaseshifts were nearly independent of stimulus intensity. The phase directionality of eardrum vibrations was smaller than that of the nerve fibers. Hence, the strong directional phaseshifts shown by the nerve fibers probably reflect the directional characteristics of extratympanic pathways. Accepted: 23 November 1996     

AN ANATOMICAL STUDY OF THE ABDOMINAL NERVOUS AND MUSCULAR SYSTEMS OF DRAGONFLY (AESHNIDAE) NYMPHS

The musculature of the fourth to eighth abdominal segments is typically composed of twenty pairs of segmental muscles associated with the body wall. In the first to third and ninth and tenth segments certain modifications to the basic plan occur in association with the abdominal-thoracic junction, the respiratory apparatus and the anal appendages. In some segments there are also paired muscles associated with the alimentary canal. Two large transverse muscles are present in the abdomen. There are eight abdominal ganglia, the first seven of which each give rise to three pairs of lateral nerves, the eighth to five pairs. In addition there are ten median abdominal nerves. The innervation fields of the various nerves are described. The first three pairs of lateral nerves of the last ganglion are homologous with the lateral nerves of the other abdominal ganglia; the fourth pair innervates most of segment nine; and the fifth pair innervates the remainder of segment nine, segment ten and the anal appendages. Certain of the abdominal muscles are innervated by branches from two different nerve roots. In segments six and seven the anterior point of attachment of the longitudinal stretch receptors is normally different from that in the other abdominal segments. This is discussed in the light of the types of movement which involve the abdomen and it seems apparent that these receptors are affected not only by swimming and abdominal flexion, as are the other longitudinal stretch receptors, but also by respiratory movements. Two distinct types of epidermal sensilla are present on the abdomen, spines and hairs. The former are the more numerous on the body, the latter on the anal appendages.     

Excitation of the common inhibitory motor neuron: A possible role in the startle reflex of the cockroach,Periplaneta americana

The responses of the widespread common inhibitory motor neuron (CI) to tactile stimulation of the cercus and the abdomen and electrical stimulation of the cercal nerve and the abdominal connectives are investigated. Tactile stimulation produces high frequency (>500 impulses/s) spike discharge in CI with the onset of CI activity preceding the discharge of the excitatory motor neurons. Electrical stimulation of the connectives demonstrates a monosynaptic connection between at least one intermediate sized fiber (conduction velocity = 3.7 m/s) in the abdominal connective and the ipsilateral CIs in the meso- and metathoracic ganglia. Electrical stimulation of the cercal nerve suggests a disynaptic path from cercal nerve to CI. Arguments are presented for a cercal afferent-to-CI reflex and the possible functional role of early excitation of CI is discussed.     

Persistent larval sensory neurons in adult Drosophila melanogaster

Using a combination of lineage tracing and laser ablation, we have identified a segmentally repeated array of embryonically produced sensory neurons that persist through metamorphosis into adult stages of Drosophila development. The persistent sensory neurons are found in all unfused abdominal segments, but there is segment-specific variation in the number of neurons observed. There are 12 persistent neurons in the first abdominal segment (A1), 18 in the second (A2), and 16 in segments A3-A7. Most are internal sensory neurons (dendritic arborization neurons and bipolar dendrite neurons), but two are associated with external sensilla on the sternite. All of these neurons and their axons define specific adult sensory pathways in the periphery and their locations and persistence through metamorphosis suggest a role in guiding the growth of adult sensory and motor axons.     

Wind-sensitive interneurones in the spider CNS (Cupiennius salei ): directional information processing of sensory inputs from trichobothria on the walking legs

Spiders can use air particle movements to localize moving prey. We studied the responses of 32 wind-sensitive interneurones in the hunting spider Cupiennius salei to prey stimuli. Stimulation with a tethered flying fly or with artificial air pulses activated plurisegmental interneurones that responded to changes in air movement velocity and were thus well suited to represent the highly fluctuating air stream typical of prey stimuli. In most interneurones (n = 18) the responses to the stimulation of different legs were not significantly different from each other. Different interneurones had different response characteristics and their latencies largely overlapped suggesting that there is parallel processing of the signals by populations of interneurones with different response characteristics. In two interneurones the number of spikes and the spiking pattern elicited by stimulation of each of the eight legs markedly differed depending on the leg stimulated. These neurones may play an important role in directional information processing. Stimulation of the adjacent legs from front to back or from back to front revealed two interneurones sensitive to the direction of successive stimulation of the legs. These neurones may be able to detect the motion of an air movement source in a preferred direction and thus act as nearfield motion detectors to localize a moving prey item. Accepted: 28 September 1996     

Identified descending brain neurons control different stridulatory motor patterns in an acridid grasshopper

During courtship sequences male grasshoppers of the species Omocestus viridulus successively perform with their hindlegs three different stridulatory movement patterns: ordinary stridulation, hindleg shaking and precopulatory movements. Microinjection of acetylcholine into protocerebral neuropil regions can either elicit complete courtship sequences or evoke one of the three motor patterns. Intracellular recordings and stainings revealed three types of descending brain neurons: B-DC-3, B-DC-4 and B-DC-5. All three types of interneurons have a medial axon position in the connectives. They cross the midline of the protocerebrum and exhibit a profuse arborization pattern within the medial dorsal protocerebral neuropil. Stimulation of each type of interneuron specifically elicits one particular motor pattern of courtship behaviour. Courtship of the grasshopper O. viridulus may therefore be controlled by successive activation of these descending brain neurons. Accepted: 27 September 1996     

Sensory modulation and behavioral choice during feeding in the Australian frog, Cyclorana novaehollandiae

This study investigates how visual and tactile sensory information, as well as biomechanical effects due to differences in physical characteristics of the prey, influence feeding behavior in the frog Cyclorana novaehollandiae. Video motion analysis was used to quantify movement patterns produced when feeding on five prey types (termites, waxworms, crickets, mice and earthworms). Twelve kinematic variables differed significantly among prey types, and twelve variables were correlated with prey characteristics (including mass, length, height and velocity of movement). Results indicate that C.␣novaehollandiae uses a different strategy to capture each prey type. Visual assessment of prey characteristics appeared to be more important in modulating feeding behavior than tactile cues or biomechanical effects. We propose a hierarchical hypothesis of behavioral choice, in which decisions are based primarily on visual analysis of prey characteristics. In this model, the frogs first choose between jaw prehension and tongue prehension based on prey size. If they have chosen jaw prehension, they next choose between upward or downward head rotation based on length and height of the prey. If they have chosen tongue prehension, they next choose between behavior for fast and slow prey. Final decisions may be the result of behavioral fine tuning based on tactile feedback. Accepted: 5 August 1996     

An endogenous elevation of cGMP increases the excitability of identified insect neurosecretory cells

In the moth, Manduca sexta, the neuropeptide, eclosion hormone, triggers a dramatic rise in the levels of intracellular cGMP within a group of 50 neurons. The cells within this group include the segmentally repeated neurosecretory cell, Cell 27. In this study the effect of cGMP on the excitability of Cell 27 was investigated using intracellular recordings. Prior to its normal elevation in cGMP, Cell 27 exhibited a high spike threshold, but this was lowered dramatically when intracellular cGMP levels increased. The latter was also associated with spontaneous action potentials. This change in excitability did not correspond with changes in either resting potential, input resistance, or action potential amplitude. A similar lowering of threshold was induced by perfusion of 8-bromo-cGMP, whereas 8-bromo-cAMP caused the threshold to increase. Intracellular recordings using various ion substitution paradigms and channel blockers provided evidence which suggests indirectly that Ca²⁺ is mostly responsible for the depolarizing phase of the action potential while a Ca²⁺-activated K⁺ current contributes to the hyperpolarization. The results of these manipulations are consistent with the hypothesis that cGMP may partially increase excitability in Cell 27 by enhancing an inward Ca²⁺ current. Accepted: 31 October 1996     

Reflex activation of locust flight motoneurones by proprioceptors responsive to muscle contractions

 This report investigates the reflex activation of locust flight motoneurones following their spiking activity. As shown elsewhere, an electrical stimulus applied to a flight muscle produces multiple waves of delayed excitation in wing elevator and depressor motoneurones. Nerve ablation experiments show that this response is initiated by the mechanical movement of the stimulated muscle, and not the antidromic spike evoked in the motoneurone. The delayed excitation still occurs in the absence of inputs from the wing receptor systems, and also when all other sources of afferent feedback are abolished, excepting thoracic nerve 2. Following complete deafferentation, spikes in flight motoneurones had no influence on other flight motoneurones. Numerous afferents in the purely sensory nerve 2 are excited by flight muscle contractions. The responses are consistent for repeated contractions of the same muscle, but differ when other muscles are stimulated. During tethered flight, changes in the activation of single flight muscles are reflected in changes of the nerve 2 discharge pattern. Electrical stimulation of this nerve causes delayed excitation of flight motoneurones, and can initiate flight activity. It is suggested that internal proprioceptors, such as those associated with nerve 2, will contribute to shaping the final motor output for flight behaviour. Accepted: 24 April 1996