Hormonal Release of Stereotyped Motor Programmes from the Isolated Nervous System of the Cecropia Silkmoth

Hormones serve as potent agents in the control of behaviour of both vertebrates (Davidson & Levine, 1972) and invertebrates (Truman & Riddiford, 1977)-Among the diverse behavioural effects of these agents, the most dramatic is their role in releasing specific stereotyped behaviour patterns. Some examples of this effect include the induction of egg-laying behaviour in the molluscs, Aplysia(Kupfermann, 1967; Arch & Smock, 1977) and Pleurobranchaea(Davis, Mpitsos & Pinneo, 1974), release of copulatory behaviour in male cockroaches (Milburn & Roeder, 1962), and the stimulation of eclosion behaviour in moths (Truman, 1973 a).

In the last example, a peptide hormone is released from the brain of the fully developed moth at the end of metamorphosis. This ‘eclosion hormone’ acts back on the nervous system to activate many of the adult behaviour patterns (Truman, 1976) and also to release a long, pre-patterned neural programme that culminates in eclosion. This neural programme involves primarily abdominal movements and can be induced by the hormone from isolated abdomens (Truman, 1971) or from the deafferented abdominal CNS in situ(Truman & Sokolove, 1972). The present report describes a completely isolated CNS preparation that responds to the eclosion hormone by the generation of the appropriate neural programme. The organization of the behaviour in the CNS and the hormonal requirements for its release are also examined.

Hyalophora cecropia were either wild-collected animals purchased as pupae from dealers or progeny from wild stock reared on leaves or on artificial diet (Riddiford, 1968). Diapausing pupae in cocoons were stored at 5 °C until needed. After they had been chilled for at least 12 weeks, pupae began adult development shortly after transfer to 25 °C in a 17L:7D photoperiod. At the end of this developmental period, the fully formed moths were found encased (‘pharate’, Hinton, 1946) inside their old pupal cuticle. Pharate moths that were due to emerge on a given day were selected by taking animals whose adult cuticles were dry because of the resorption of their moulting fluid. On each day, selection was made just prior to lights-on because eclosion hormone release in H. cecropia begins just after this signal. The abdomen of each animal was then isolated by clamping with a haemostat between the thorax and abdomen and discarding the portion anterior to the clamp. Abdomens isolated in this manner respond to an injection of the eclosion hormone with a stereotyped behavioural response that is described below.

Initial attempts to devise an isolated nervous system preparation involved pinning the CNS in a chamber and perfusing the preparation with oxygenated saline. These nervous systems continued to show spontaneous, tonic motor activity for hours, but they showed no stereotyped response to the eclosion hormone. Behavioural responses to the hormone were only obtained when air was supplied to the CNS through its normal route, i.e. through the tracheal system.

The old pupal cuticle was removed from around the isolated abdomens. Each abdomen was opened anteriorly, the intact midgut and rectal sac removed, and the abdominal cavity flushed with a saline (4 mm-NaCl, 40 mm-KCl, 18mm-MgCl₂, 3 mm-CaCl₂, 150 mm glucose; adjusted to pH 6·5 in 2·5 mm-KH₂PO₄-KHCO₃ buffer) modified from that described by Weevers (1966). The ventral cuticle and epidermis were removed taking care not to wet the external spiracular areas. The abdominal nervous system of the pharate adult consists of a chain of three simple ganglia (A₃, A₄, A₅), and a compound terminal ganglion (A₆). Each ganglion is supplied by bilateral tracheal trunks that lead from the paired spiracles. Each pair of trunks is also interconnected by a ganglion by-pass branch that was severed in these preparations. The abdominal CNS and its attached tracheal supply including the spiracles were dissected from the abdomen and pinned in a paraffin dish. One tracheal trunk leading to each ganglion was severed near the spiracle and fitted over a glass cannula ; the spiracular valve of the paired trunk was then removed so that the hydrophobic lumen of the tracheal cluster opened to the air and floated on the surface of the bath (Fig. 1). Cannulae were fitted into the tracheal trunks to ganglia A₃, A₄ and A₅. The CNS was then ventilated by means of a peristaltic pump which forced air through the cannulae. Isolated preparations were slowly perfused with saline, or fresh saline was periodically added to the bath.

The motor activity of the isolated nervous system was monitored by two to four suction electrodes placed on the stumps of the segmental dorsal nerves. These nerves supply the longitudinal intersegmental muscles (Libby, 1961) that are arranged around the circumference of the fourth to sixth abdominal segments and which are responsible for most of the behaviour displayed by the abdomen. Spontaneous motor activity was monitored on a Tektronix 4-beam oscilloscope and recorded on a Hewlett-Packard 4 channel Instrumentation tape recorder. Simultaneously, the motor activity from each root was processed through a pulse rate integrator and the integrated output displayed on a Beckman Dynograph recorder. During later analysis of the motor activity, the data on tape were replayed at four times the recording speed through the same integrating circuits.

The behaviour of intact H. cecropia and of isolated abdomens was monitored as described in Truman (1971). A thread was waxed to the tip of the abdomen and Attached to one arm of a lever which wrote on a revolving smoked drum, thus monitor-ing the frequency of abdominal movements. The type of movement was indicated by the length and direction of the pen excursion and also checked by observation.

Emergence from the pupal skin is accomplished through rhythmic eclosion movements which take the form of peristaltic waves of strong bilateral contractions progressing segmentally up the abdomen (Fig. 2). Arrival of the wave at the thorax is followed by vigorous shrugging movements of the wing bases. In H. cecropia and other satumiid moths, these eclosion movements are preceded by a series of stereotyped abdominal movements which involve primarily abdominal rotations. In the original description of this behaviour in H. cecropia, the term ‘pre-eclosion behaviour’ was used to denote the abdominal movements which were displayed immediately prior to eclosion (Truman, 1971). In subsequent reports, however (e.g Truman & Sokolove, 1972; Truman, 1973a) the term was used in a looser sense: the whole of the behavioural sequence including eclosion, was described as ‘pre-eclosion behaviour’. The results in the present paper argue that the pre-eclosion and eclosion behaviours are distinct behavioural units and, consequently, I have returned to the more restrictive use of the two terms.

The eclosion hormone was obtained from the brain neurohaemal organs (the corpora cardiaca) dissected from pharate Manduca sexta moths. The glands were homogenized in a small volume of saline in a micro-homogenizer. Unless otherwise stated, isolated abdomens received approximately one pair of glands per abdomen. In the case of the isolated CNS preparations, the hormone extract was dropped on the CNS in an amount that would give a final concentration of approximately two corpora cardiaca pairs per ml.

In Hyalophora cecropia the pre-eclosion behaviour consists of an initial active period followed by a period of relative or complete quiescence. This is then followed by the eclosion behaviour with its highly stereotyped eclosion movements. Fig. 3 gives examples of kymograph records from emerging H. cecropia moths. Of the 45 records that were randomly selected for analysis, 43 of the animals showed a typical pre-eclosion behaviour (e.g. Fig. 3 a−h). One of the remaining two moths showed no pre-eclosion behaviour before emergence whereas the other performed a behaviour that lacked the quiet period (Fig. 3i). Of the animals that gave a typical behaviour, the active period had an average duration of 32 ± 8 min and included an average of 26 discrete bouts of activity (range 7−47). The movements seen during these bouts were primarily rotary motions of the abdomen but various types of abdominal flexions and twitches also occurred. In many moths the frequency of activity bouts peaked about midway through the active period. This period was followed by a period of relative quiescence which lasted 31 ± 8 min. The level of activity during the quiet period varied from a moderate drop in the frequency of movements (Fig. 3 h) to almost complete inactivity (Fig. 3b, e). The transition from the active to the quiet period was usually rather abrupt. In the cases in which it was gradual, the transition was judged to occur at the point at which the frequency dropped down to the level characteristic of the subsequent quiet period.

The eclosion behaviour began at the end of the quiet period. The characteristic eclosion movements (Fig. 2) occurred at a frequency of about 3−5 per min. Rotary movements were also occasionally observed during this period.

The isolated abdominal nervous system of H. cecropia was set up as described in Methods and spontaneous motor activity was monitored. In the absence of the eclosion hormone, the spontaneous activity typically consisted of a low level of tonic firing by one or two motor units in each dorsal nerve. In some preparations, this tonic activity was occasionally interrupted by spontaneous motor bursts. In six preparations that were not exposed to hormone, this low level of motor activity continued essentially unchanged through the two-to four-hour recording period.

Fig. 4 is a record of the spontaneous motor activity from an isolated CNS preparation. During the first 80 min no hormone was in the bath, and the motor activity consisted of the firing of one or two tonic units with infrequent bursts by many motoneurones. An eclosion hormone extract was then added to the bath. Fifty min later, the CNS began a period of frequent bursting. After about 35 min, the frequency of bursting declined and the nervous system became relatively quiet. This quiet period was then abruptly terminated by the resumption of bursting. After the onset of this last period, bursts occurred at regular intervals for a number of hours. Of 18 isolated CNS preparations that were exposed to the eclosion hormone, 16 responded similarly to the preparation shown in Fig. 4. After an average latency of 44 ± 20 min, the nervous systems showed an initial period of frequent bursting, followed by a quiet period, and then the resumption of bursting.

The isolated CNS preparations showed an average of 29 bursts (range 9-74) during the first active period. The most commonly seen burst had a rotary patterning (Fig. 5). Each burst lasted up to 30 s or longer and showed repeating volleys of motoneurons firing. Volleys occurred simultaneously in ipsilateral roots along the chain of ganglia but there was strong right-left alternation. This alternation presumably results in the rotary movements of the abdomen which are seen in intact animals. Although sequential synchrony and bilateral alternation were consistent features of these bursts, the finer details of their patterning varied greatly, even from burst to burst in the same preparation. For example, Fig. 5 shows four consecutive rotary bursts recorded from one preparation. The period of alternation ranged from 1·2 to 9·6 s. In addition to these presumptive rotary bursts, all roots sometimes fired in synchrony. This type of burst presumably corresponds to the occasional twitches and flexions that are observed during the pre-eclosion behaviour of intact animals.

The onset of the quiet period was characterized by a marked decline in burst frequency. Bursts which occurred during this time typically showed the same rotary patterning as those of the preceding period. A low level of tonic motor activity usually persisted throughout the quiet period, but in some preparations it disappeared completely.

The quiet period was terminated by the resumption of frequent bursting. But instead of being rotary bursts, the new motor bursts characteristically showed the eclosion patterning (Fig. 6). During a burst, the frequency of motoneurone firing rose to a peak and then abruptly terminated. Within each ganglion the motor output was bilaterally symmetrical, but the firing in the A₆ roots would terminate, followed 1−2 s later by the major burst and termination in A₅, and then in A₄,1−2 s after that. At the end of the A₄ burst, the posterior roots sometimes showed a rebound in activity (Fig. 6 b). Eclosion bursts typically occurred at regular intervals for a number of hours. Bursts having a rotational patterning were also occasionally seen early in the eclosion phase.

The results from the completely isolated CNS are similar to those obtained from the deafferented abdominal nervous system in situ(Truman & Sokolove, 1972). When exposed to the eclosion hormone, the isolated nervous system performed the motor programme for the pre-eclosion and eclosion behaviours. Thus, the information for both behaviours is patterned into the abdominal CNS, and both are triggered in their proper sequence by the direct action of the hormone on the CNS.

Fig. 7 presents examples of the motor activity evoked by the eclosion hormone from isolated CNS preparations. The nine records were selected to show the range in response that was observed. One striking feature was the variable duration of the pre-eclosion behaviour. The behaviour lasted for 57−325 min with an average duration of 123 min. The length of this behaviour appeared to be related to the ‘general excitatory state’ of the nervous system. Preparations showing spontaneous motor bursts prior to hormone addition performed behaviours that were shorter than those displayed by nervous systems showing only tonic activity (Table 1). Also, the latency between hormone addition and the start of the pre-eclosion behaviour was shorter in these spontaneously bursting cords.

Of the 16 preparations that responded to the hormone, two showed no discernible quiet period (e.g. Fig. 7i) and eclosion bursts began abruptly after a long series of rotary bursts. In the remaining 14 cases, a quiet period was evident. During this time the CNS usually showed only occasional bursts but in two examples (e.g. Fig. 7 h), the quiet period involved simply a small decline in the burst frequency. In addition, two preparations showed one and two peristaltic bursts respectively in the middle of the quiet period.

The onset of the eclosion behaviour was usually marked by an eclosion burst (Fig. 7 a, b, d, g, h) but rotational bursts were also occasionally seen during the initial stages of this period. The occurrence of rotational bursts became progressively rarer as the eclosion phase continued. The likelihood that a rotary burst would appear among the eclosion bursts was related to the length of the preceding pre-eclosion behaviour. All of those preparations that showed rotational bursts during the eclosion period (N = 8) had a pre-eclosion behaviour that lasted less than no min. When the pre-eclosion behaviour was longer than 110 min, no rotary bursts were subsequently seen during the eclosion period. The significance of this difference will be discussed below.

In most cases at the first peristaltic burst of the eclosion period, all of the motor roots showed the phase relationships that were characteristic of eclosion bursts. However, in three preparations, the eclosion period began with bursts that showed abnormal patterning in that all of the dorsal nerves terminated activity essentially in synchrony (Fig. 8). But after the first few bursts, the roots abruptly switched into their characteristic phase relationships. Once the proper pattern was established, the bursts showed little change in phase during the remainder of the recording sessions. Also, in some instances, the first few eclosion bursts were smaller than were the later bursts.

As described above, the hormonally induced motor activity generated by the isolated CNS showed considerable variation in a number of its temporal characteristics. These variations were used to deduce some aspects of the organization of the behavioural programmes. One variable feature was the duration of the pre-eclosion behaviour. In intact animals, this behaviour lasted about 60 min and was evenly divided between the active and the quiet periods. In the isolated CNS, the duration of the pre-eclosion behaviour was generally longer than that seen in intact animals. As seen in Fig. 9 a, under these conditions, the length of the active period was highly correlated with the length of the quiet period (r = 0·83 ; P < 0·01) and they varied in essentially a one to one manner. Therefore, the increase in the duration of the preeclosion behaviour occurred through the proportionate lengthening of both the active and the quiet periods.

The increase in the length of the active period could be due to the occurrence of additional bursts, an increase in the interburst interval, or both of the above. Fig. 9b shows that the length of the active period varied directly with the average duration of the interburst interval (r = 0·77; P < 0·01). Also, the average number of bouts of activity shown during the active period of intact animals was similar to the average number of motor bursts generated by the isolated CNS: 26 and 29 respectively. Consequently, the lengthening of the active period in the isolated CNS appears to come about through an increase in the interval between bursts and not by the addition of new bursts.

From the above, one can conclude that when the pre-eclosion behaviour is slowed in the isolated preparations, all of the temporal parameters of this behaviour (the burst frequency during the active period as well as the length of the active and quiet periods) are proportionally affected. Therefore, these parts of the pre-eclosion behaviour appear to make up a discrete behavioural unit.

In order to determine if the eclosion behaviour was also part of the pre-eclosion behaviour unit, the temporal characteristics of this behaviour were examined. Fig. 9c compares the duration of the pre-eclosion behaviour with the average interburst interval during eclosion. No significant correlation was found between these two parameters (r = 0·11). When the frequency of eclosion bursts was compared with the other temporal parameters of the pre-eclosion behaviour, there was also no correlation. Consequently, the timing of motor activity during the pre-eclosion and eclosion behaviours appears to be controlled separately and these behaviours may represent distinct behavioural units.

In the isolated CNS, both the pre-eclosion and eclosion behaviours show wide variations in the frequency of bursting. It was of interest to determine if these changes in frequency affected the length of the respective motor bursts. Fig. 10 compares the average interburst interval during the eclosion behaviour with the average duration and average phase delay of the respective eclosion bursts. In both cases, there was no correlation between the temporal structure of the bursts and the interval between successive bursts (r = −0·28 and −0·03 respectively). Consequently, the form of the eclosion bursts appears to be independent of the frequency of bursting during the eclosion period.

In the case of the pre-eclosion behaviour, the rotary bursts are made up of repetitive motoneurone volleys which presumably produce bouts of repetitive abdominal rotations. The length of these volleys usually varied so widely between rotary bursts (see Fig. 5) that the determination of average values for each preparation appeared meaningless. Consequently, for a few preparations the average length of the volleys comprising each rotary burst was computed and compared with the interval between successive rotary bursts. Fig. 10c, d shows an example of one such analysis. There was no correlation between the length of the volleys within a rotary burst and the interval either before or after that burst (r = −0·20 and −0·28 respectively).

The above records were obtained from preparations that were continuously exposed to extracts containing the eclosion hormone. It was of interest to determine whether this continuous exposure to hormone was essential for the display of the preeclosion and eclosion behaviours or whether a short initial exposure would suffice. Eclosion hormone was added to isolated CNS preparations, and at various times thereafter the hormone was washed out by 10−15 changes of the bathing medium over a 2 min period. The effects of various lengths of exposure to the eclosion hormone are summarized in Table 2. The wash-out procedure itself was ineffective in eliciting the motor programmes. With exposures of 5 min or less, the preparations were erratic in their responses. Each of the three groups (1, 2·5, and 5 min) included one preparation which showed no motor response to the treatment. These preparations were later re-exposed to the eclosion hormone and they all subsequently performed the appropriate behaviours, thereby indicating that they were competent to respond but that the initial exposure was inadequate to trigger the behaviours. Of the six nervous systems that did respond, three (Figs. 11, 12 a, c) displayed a pre-eclosion behaviour that was not obviously different from preparations that had had tonic exposure to hormone. Of the three preparations which showed abnormal behavioural responses, two deserve special attention. In these two cases, the duration of the pre-eclosion behaviour was very short (about 40 min) and a quiet period was absent. Also, with the onset of the eclosion behaviour the eclosion bursts were abnormal in that motor activity in all dorsal nerves terminated in synchrony. Although this abnormal eclosion pattern was occasionally seen in normal preparations (Fig. 8), these two preparations showed the abnormal patterning for exceedingly long periods; in the case of Fig. 12e, more than 2 h elapsed before the proper phase relationships were finally established. The significance of these aberrant patterns will be considered below.

Five preparations were exposed to the hormone for 10−15 min. All of these showed the pre-eclosion behaviour followed by eclosion (e.g. Fig. 12f g, h). Also, in all instances the organization of the pre-eclosion behaviour appeared to be within the normal range of variation. Consequently, one can conclude that the activation of the pre-eclosion and eclosion behaviours can be accomplished by a brief exposure to hormone. After this brief exposure, removal of hormone from the bath does not prevent the initiation and performance of the two behaviours.

In silkmoths, sensitivity to the eclosion hormone is acquired very late in adult development (Truman, 1976), and it is then lost after the eclosion hormone causes the moth to emerge (Truman, unpublished observations). To determine how long this sensitivity would persist in abdomens that had been isolated from pharate moths but not exposed to hormone, abdomens were challenged with a standard dose of hormone at various times after isolation. Abdomens that were injected with hormone 1 day after isolation responded by showing both the pre-eclosion and eclosion behaviours (Table 3). Those injected after 2 days showed a mixed response: some did not respond at all, others showed both behaviours in proper sequence, and still others showed only the peristaltic movements characteristic of the eclosion behaviour. Importantly, in these last cases, this eclosion behaviour did not begin until 70−90 min after hormone injection-at the time it should have appeared if the pre-eclosion behaviour had also been displayed. By 3 days after isolation, abdomens injected with eclosion hormone typically did not show either behaviour. Thus, as a pharate abdomen ages, it progressively loses its ability to respond behaviourally to the eclosion hormone. This loss is not due to the abdomens becoming moribund because such preparations typically remain alive and sensitive to tactile stimulation for 2 weeks or longer. Moreover it occurs in a step-wise fashion with first the disappearance of the pre-eclosion behaviour and then of the eclosion behaviour.

After abdomens are isolated from pharate H. cecropia moths, some of the abdominal, intersegmental muscles begin to degenerate as the abdomens age. Therefore, it was possible that the disappearance of the pre-eclosion behaviour might be related to muscle degeneration rather than to changes in the CNS. This hypothesis appeared unlikely because 2-day-old abdomens that did not show the pre-eclosion behaviour nevertheless showed rotary movements in response to tactile stimulation. Also their intersegmental muscles showed some signs of degeneration but they were still contractile. Conclusive evidence in favour of a central change was the fact that an isolated CNS taken from a 2-day-old abdomen also showed the selective loss of the preeclosion behaviour. As seen in Fig. 13, in this preparation the pre-eclosion behaviour was absent (or at best represented by one rotary burst that occurred 48 min after hormone addition). The eclosion behaviour, by contrast, began abruptly 202 min after hormone addition and showed frequent, strong eclosion bursts. Consequently, the loss of the pre-eclosion behaviour represents a loss in the central motor programme. Also, this behaviour can be lost without a significant effect on the timing of the eclosion behaviour.

In moths the onset of the pre-eclosion and eclosion behaviours is correlated with the appearance of eclosion hormone in the blood (Truman, 1973b; Reynolds, Taghert & Truman, in preparation). Although this hormone normally releases these behaviours, it is not absolutely required for their performance. This conclusion is based on the fact that debrained silkmoths can nevertheless emerge (Truman, 1971) even though they presumably lack the eclosion hormone. But it is important that they emerge a day or two later than sham-operated controls and that they usually show an eclosion behaviour without the initial pre-eclosion behaviour (Truman, 1971). Also, there were two examples of brainless moths that performed the pre-eclosion behaviour but not eclosion. Thus, it appears that in brainless moths, the centres controlling the preeclosion and eclosion behaviours can spontaneously become active and can do so independently of one another. Under normal conditions, the hormone apparently serves to trigger the co-ordinated activation of these behavioural programmes.

The above conclusions are supported by the results of the wash-out experiments. The activation of both the pre-eclosion and eclosion behaviours was accomplished by the brief presence of the hormone in the bathing medium around the CNS. In the case of very brief exposures (less than 5 min) the resultant behaviours were sometimes abnormal in their temporal and spatial patterning. The same types of abnormalities were also produced by selectively exposing parts of the abdominal nervous system to hormone while the rest remained unexposed (Truman, in preparation). It is likely that the aberrant behaviours that are seen during very short hormone exposures occur when some parts of the nervous system have sufficient contact with the hormone to become activated whereas other parts do not.

The wash-out experiments indicate that the eclosion hormone acts as a trigger to cause the release of the appropriate behaviours. The hormone presumably binds to receptors in the nervous system. This binding then effects long-term changes in the CNS which eventually result in the performance of the pre-eclosion and eclosion behaviours even though the hormone itself may no longer be present. However, it could be argued that the eclosion hormone might bind very tightly to its receptors such that it cannot be removed by the washing procedure. Consequently, the hormone could be present bound to its receptors throughout the behavioural sequence. Alternatively, there may be accumulation of hormone within the neural sheath such that it cannot be readily washed out. These last possibilities cannot be excluded at present.

Indirect support for the triggering function of the hormone comes from studies of the variations in cyclic AMP concentration in the abdominal CNS of Hyalophora cecropia(Truman, Fallon & Wyatt, 1976). After hormone administration, cyclic AMP reaches a peak concentration before the onset of the pre-eclosion behaviour and then declines back to basal levels by the middle of the active period. This early rise and fall in cyclic AMP concentration may indicate that the hormone is acting only during this early period.

Studies on a number of animals have demonstrated that simple repetitive motor patterns such as flight in locusts (Wilson, 1961) and walking in cockroaches (Pearson & Iles, 1970) arise from pre-patterned information within the CNS. Similarly, certain rapid escape responses such as escape swimming in the nudibranch Tritonia(Dorsett, Willows & Hoyle, 1969) are also due to pre-patterned neural information. The basic motor patterns involved in these behaviours have durations of the order of milliseconds to seconds. With longer term behaviours, there is usually a more complex arrangement of motor activity into a definite temporal ‘score’ that may last for minutes or hours. During the progression of these behaviours, discrete motor patterns may be recruited, displayed for a few minutes and then disappear with the subsequent addition of new patterns. Two examples of an arrangement of simple motor acts into a complex behavioural programme are the courtship behaviours of some insects (Loher & Huber, 1966) and the ecdysial behaviours of crickets (Carlson & Bentley, 1977) and moths (Truman, 1971; Truman & Sokolove, 1972). In these cases, the information not only for the simple motor patterns, but also for the temporal arrangement of these patterns into a complex behavioural programme, is apparently built into the CNS of the animal.

In moths, a hormone directly triggers two behavioural units-the pre-eclosion and the eclosion behaviours. Each unit appears to have a hierarchical organization that includes one or more motor pattern generators to produce the characteristic movements (i.e. rotation, flexion, peristalsis) as well as a ‘burst timer’ that provides the temporal information to arrange these discrete motor patterns into a complex behavioural programme. In both cases, the burst timer has a more or less rhythmical output. In many kinds of rhythmically occurring behaviours, such as walking in cockroaches (Pearson & Iles, 1970), the duration of at least part of the basic motor pattern is strongly influenced by the frequency of the controlling oscillator; an increase in frequency causes a compression of the motor pattern whereas a decrease in frequency results in its expansion. Thus, in such cases, the oscillator itself may be an inherent part of the pattern-generating system. In other cases, such as the chirp rhythm of crickets (Bentley, 1977), the form of the chirp (i.e. the number of pulses and interval between pulses) is independent of the chirp frequency. Consequently, it is likely that these two features of cricket song are functionally distinct. The rhythmically occurring eclosion behaviour is similar to the cricket system in that the duration of the eclosion burst is independent of the frequency of the bursting (Fig. 10a, b). Consequently, the pattern generator for eclosion bursts is apparently not a component of the eclosion burst timer but rather the two are functionally distinct entities within the CNS. Similarly, in the case of the pre-eclosion behaviour, the length of rotary burst volleys is unrelated to the interval between bursts (Fig. 10c, d). It follows that the pattern generator for rotary bursts is functionally distinct from the pre-eclosion burst timer which triggers these bursts during the active period.

In some insects there is evidence that the same neural oscillator may participate in the timing of more than one rhythmic behaviour (Miller, 1974). The timing of preeclosion and eclosion behaviour is apparently controlled by distinct burst timers that are sequentially activated rather than by the same timer whose coupling switches to different motor pattern generators. This conclusion is based on the fact that there is no correlation between the average frequency of bursting during the pre-eclosion and eclosion behaviours (r = 0·08). Also, the eclosion burst timer acts like an oscillator with a relatively stable frequency. By contrast, the frequency of the pre-eclosion burst timer is less stable and it tends to rise and then decline during the course of the active period. Finally, in 2-day-old abdomens, the pre-eclosion behaviour is lost without significant interference with the timing of the eclosion behaviour (Fig. 13). Since stimulation of these abdomens will nevertheless evoke rotary movements, the disappearance of the behaviour is apparently due to a disruption, not of the rotary pattern generator, but of the pre-eclosion burst timer. Since this burst timer can apparently be lost without influencing the one that controls eclosion, it is most likely that the two burst timers are functionally separate.

It would seem, therefore, that the pre-eclosion and eclosion behaviours are separate behavioural units that are comprised of distinct burst timers and motor pattern generators. The organization of the eclosion behaviour is relatively simple, consisting of a burst timer, which appears to be an oscillator, and an eclosion pattern generator. That of the pre-eclosion behaviour is more complex and can be divided into two subunits-the active and the quiet periods. The active period has been considered above, but what is the origin of the quiet period ? This period might simply represent a waiting time between the shutting-off of the pre-eclosion burst timer and the activation of the eclosion oscillator, or it may be a programmed period of inhibition of motor activity. The behaviour of the isolated CNS preparations indicate that the latter possibility is more likely. The fact that the length of the quiet period varies in an approximately one-to-one relationship with the duration of the active period implies that both are programmed parts of the same behavioural unit and that the times of termination as well as initiation of both halves are essential parts of this programme. Presumably then, the quiet period represents an active inhibition of motor activity and the various levels of activity seen during this period reflect individual variation in the strength of this inhibition. Indeed, in some preparations, the presumed inhibition was strong enough to repress even the low level of tonic motor activity that is seen in an unstimulated nervous system. The reappearance of a few rotational bursts after the quiet period in some of the preparations may also indicate a release from inhibition.

What is the relationship of the eclosion hormone to these two behavioural units Does the hormone release the first unit and the completion of that in turn triggers the next, or are both behaviours directly triggered by the hormone ? When a threshold dose of eclosion hormone is injected into an isolated H. cecropia abdomen, one occasionally sees an abdomen that performs the pre-eclosion behaviour but fails to perform the eclosion behaviour (Reynolds & Truman, in preparation). From this result, it is clear that the eclosion behaviour is not triggered simply by the termination of the pre-eclosion behaviour. By contrast, the injection of hormone into 2-day-old abdomens (Fig. 13) causes the display of only the eclosion behaviour but at a time that would be expected if the preceding behaviour had also been performed. Consequently, it appears as if both behaviours are directly released by the eclosion hormone and that their sequential order of occurrence is primarily a function of the different response latencies of the respective neural programmes.

Even though the sequential activation of the two behaviours reflects mainly their different latencies, there is probably also some interaction between the two behavioural programmes. In the isolated preparations, when the duration of the preeclosion behaviour was short, rotational bursts were often seen early in the eclosion period, but when the duration was long, only peristaltic bursts were seen after the quiet period. In the shorter duration behaviours, it may be that the quiet period ended before the time that all of the eclosion unit was fully activated. Also, the two cases in which isolated eclosion bursts were observed in the quiet period (e.g. Fig. 7f) may be examples of the reverse situation. Finally, preparations such as that in Fig. 7 h, in which occasional rotary bursts occur during the eclosion behaviour in the place of expected eclosion bursts, further suggest some interaction between the two behavioural units.

The above conclusions on the organization of the pre-eclosion and eclosion behaviours are presented in Fig. 14. Each behaviour comprises a distinct behavioural unit. Each has a two-tier hierarchical arrangement with a burst timer serving to drive separate motor pattern generators. This organization is very similar to that proposed by Carlson & Bentley (1977) for the organization of ecdysial behaviour in crickets. In the case of the moth, both programmes are activated simultaneously and in parallel, but they show different latencies. The quiet period represents a programmed period of inhibition which causes a marked decline in the frequency of rotary movements. Whether the inhibition is at the level of the pre-eclosion burst timer or the rotary pattern generator is unknown. The quiet period may also serve to inhibit the eclosion behaviour in cases in which the eclosion unit is activated early. The level of this inhibition is also unknown.

In silkmoths the eclosion behaviour serves to extract the moths from the pupal cuticle. The pre-eclosion behaviour is presumably homologous to the preparatory behaviours shown by some other insects (e.g. Carlson & Bentley, 1977), but its function in moths is unknown. The central programmes underlying these two behaviours show varying degrees of dependence on other central and peripheral factors. The importance of these factors can best be seen by comparing the behaviour of the isolated CNS with that of isolated abdomens and intact animals.

As described above, isolated CNS preparations have required as long as 325 min to perform the pre-eclosion behaviour, whereas in intact animals this behaviour has a duration of only about 60 min. This pronounced slowing of the pre-eclosion behaviour seems related to the central excitatory state of the isolated abdominal CNS (Table 1). The lack of input from high centres in the head or thorax does not appear to be a factor because isolated abdomens exposed to maximal doses of the eclosion hormone show pre-eclosion behaviours that have a normal duration (63±17 min, N = 11). This decline in responsiveness may be due either to the lack of sensory input or to inadequate aeration of the preparation. Since some isolated preparations do show behaviours with a duration within the normal range (e.g. Fig. 7d, h), the aeration of the preparation may be the critical factor.

Aside from a possible indirect influence through its effect on the level of general excitability, sensory input appears to have little direct effect on the temporal organization of the pre-eclosion behaviour. The level of activity during the active period varies widely in both the isolated CNS and the intact moth but the average number of bursts generated by the isolated CNS is essentially the same as the number of movements shown by intact animals. Similarly the variation in the level of activity seen during the quiet period appears quite similar under the two conditions. Therefore, it appears that the pre-eclosion burst timer may be set for a specific number of bursts and that this number is not affected by normal sensory feedback.

In contrast to the pre-eclosion burst timer, that controlling the eclosion behaviour appears to be directly influenced by sensory feedback. Eclosion movements occur at a frequency of 3−5 per min in whole moths, whereas in the isolated CNS the frequency is always much lower. A major reason for this difference is seen by considering the eclosion of isolated abdomens. Abdomens that were still covered with pupal cuticle at the onset of eclosion typically showed the normal frequency of eclosion movements. By contrast, those that had had their pupal cutical removed and a thread waxed to the tip of the abdomen to record movements often showed a lower frequency of eclosion movements. Therefore, it appears that the stimulus provided by the surrounding pupal cuticle is an important factor in determining the frequency of eclosion bursts.

Besides showing these bursts at a lower frequency, the isolated CNS may continue to generate eclosion bursts for hours. The act of shedding the pupal skin probably plays a role in the termination of eclosion behaviour, because this terminates more quickly in isolated abdomens which shed their cuticle than in peeled abdomens or the isolated CNS. But in no case does eclosion behaviour terminate as rapidly as in intact animals. Also, when abdomens were isolated from emerged moths which had terminated eclosion and had initiated wing spreading, the abdomens relapsed into showing regular eclosion movements within a few minutes after isolation. Thus, it appears that under normal conditions the eclosion behaviour is terminated by inhibition from higher centres in the CNS.

Besides its effect on the burst timers, sensory feedback may also influence the individual motor pattern generators. This influence is most marked in the case of the rotary bursts. In the isolated CNS, the period of rotation may vary erratically from burst to burst (Fig. 5). This wide variation is not seen in intact animals or in isolated abdomens and presumably indicates that sensory feedback has a major effect on the output of the rotary pattern generator. In contrast to the rotational bursts, the peristaltic bursts are highly stereotyped and show much less variation either between or within preparations. Observations on intact moths show that the eclosion wave progresses up the abdomens at a rate of approximately one segment every 1−1·5 s. A similar timing is noted in the segmental progression of eclosion bursts in the isolated CNS. Thus, sensory feedback apparently has little role in the expression of the eclosion pattern generator.

Thus, as summarized in Fig. 15, sensory feedback appears to influence the two behavioural units at different levels. The pre-eclosion behaviour is generated by a pentral motor tape (Hoyle, 1970) having a defined duration with a definite beginning and end. Conditions that lower the central excitatory state may slow the tape, but sensory feedback appears to play little or no role in influencing the number of motor commands that will be generated during the play-out of the tape. However, the output of the tape is greatly affected by sensory input.

In contrast to the pre-eclosion behaviour, the eclosion behaviour appears to be controlled by an ‘oscillator’ that has no programmed duration. Its frequency is markedly influenced by sensory input, especially that provided by pupal cuticle, and it is tumed-off by higher centres in response to the shedding of the cuticle. Although the frequency of the eclosion oscillator is influenced by sensory conditions, the eclosion pattern generator is apparently not. This presumably reflects the fact that these movements are performed for a unique reason in a very predictable environment (i.e. inside the pupal cuticle). Consequently, there is apparently little need for modifying the pattern of the behaviour.

I thank Professors L. M. Riddiford, J. Palka, Dr S. E. Reynolds and Mr P. H. Taghert for a critical reading of the manuscript. The research was supported by NSF grant PCM 75-02272 Aoi and NIH grant 1 Roi NS13079-01. The author was supported by NIH Career Development Award 1 K04 NS00193-01.