The Effect of Acoustic Signals On the Chirp Rhythm in the Bush Cricket Pholidoptera Griseoaptera

Neural rhythm generators or ‘pacemakers’ appear to play an important part in the production and timing of the chirps in crickets and bush crickets and in the control of the wing movements which produce the syllables (Huber, 1963, 1967; Jones, 1966a; Shaw, 1968). As a result of his work with gryllids and acridids, Huber has described three control centres: the mushroom bodies, the central body of the brain and the thoracic ganglia. Bentley (1969) has made recordings which indicate that there are units in the mesothoracic ganglion of G. campestris which display an oscillation of activity at the chirp rhythm. More recently, Kutsch & Otto (1972) have shown that normal chirps occur even when there is no neural connexion between head and thorax. Thus the network of thoracic neurones must provide generators for both chirp and syllable rhythms, the mushroom bodies and central body presumably acting by excitation or inhibition of this thoracic system.

In the bush cricket Pholidoptera griseoaptera, natural and artificial sound signals appear to have an inhibitory effect on the chirp rhythm generator (Jones 1966a, b). This effect is rapid and reliable (reaction time ca. 50 msec), but may be followed by excitatory or depressant after-effects which are much more variable. Similar results have been obtained with the bush cricket (‘Katydid’), Pterophylla camellifolia (Shaw, 1968), the house cricket, Acheta domesticas (Heiligenberg, 1966, 1969), and the field cricket, Gryllus campestris (Jones & Dambach, 1973).

This paper analyses the effect, on the chirp rhythm in Ph. griseoaptera, of signals given at different phases of the chirp cycle. The results indicate clearly that the generator of the chirp rhythm is reset by the signals; those given late in the cycle have the greatest effect.

Adult males were caught, kept in large ‘locust’ cages and fed on soaked wheat seeds and lettuce. Individual males were marked by painting dots on the pronotum with ‘Humbrol’ model paint. The experiments were carried out within two weeks of capture.

In preliminary experiments, the artificial sound signals were given at regular intervals, using the techniques described in an earlier paper (Jones, 1966b), and thus had no regular phase relation with the chirp cycle. In all other experiments the signals were triggered by the insect ‘s own chirp and could thus be given regularly at any predetermined time in the chirp cycle.

When the insect chirped, the sound was picked up by a microphone and recorded on one channel (ch. 1) of a stereo tape recorder (Uher 4400). The monitor output from the recorder was amplified and used to trigger a wave generator (Tektronix type 162) to give a saw-toothed wave with a slow rise time. When this wave reached a predetermined threshold, it triggered a square-wave generator (Tektronix type 161); the time the saw-toothed wave took to reach the threshold determined the chirp-to-signal period (x). The square wave was used to gate the output of a sine-wave generator in the circuit previously described (Jones 1966b). The output of this circuit was amplified and played to the insect through a calibrated electrodynamic tweeter loudspeaker (LPH 65), and was also recorded on the second channel (ch. 2) of the tape recorder. The period of the saw-toothed wave determined when the apparatus could be triggered again by the insect. This allowed the insect to chirp freely for up to 10 sec after each signal before the next could be triggered.

In all experiments, the signals were 100 msec long (chirp length = 70–100 msec) and the sound pressure level was 70dB (root mean square relative to 2·10⁻⁵N/m²). In preliminary experiments (see Fig. 2a) the frequency was 12 kHz, but in subsequent ones (in which the insect triggered the signal) was 15 kHz. Previous experiments (Jones, 1966b) have shown that the insects respond well to both frequencies. Rise and decay times between 1 and 10 msec were used. In some experiments the rise and decay times were so abrupt that the loudspeaker gave ‘clicks’; the results of these experiments could not be distinguished from those with slower rise and decay times, and so are included with them.

Each experiment was carried out during the late evening and night, when the insect was most active. A solitary male, in a wooden-framed muslin cage (10×10×10 cm), was placed in position in an open-fronted enclosure (approximately 1 m wide × m high × m deep) made of slabs of rock wool (a sound-absorbent material). The open front of this enclosure faced the loudspeaker, which was mounted on a large slab of rock wool, 2 m from the insect. The microphone was placed near the insect in a position that did not interfere with the sound field of the loudspeaker. The experiment was monitored and controlled from a distant part of the laboratory. Before and after each test with the signals, the insect was allowed to chirp freely for several minutes.

The temperature, recorded during each experiment, was between 23° and 25 °C in all experiments except a preliminary experiment (Fig. 2a) in which it was 18·5 °C.

Fig. 1 shows the effect of signals given late in the chirp cycle of a male which is chirping steadily at a high rate (ca. 50–60 per min). Normally the signal delays the next chirp and resets the chirp cycle (Fig. 1 a). After the delay, there may be a slight decrease in the period (from the beginning of one chirp to the next). Occasionally, chirp and signal may approximately synchronize (Fig. 1 b); this seems to happen when the signal is too late to inhibit the chirp (Jones, 1966a, b). In this case, the cycle is not reset.

Fig. 2a shows the results of an experiment in which the signals were not triggered by the insect, but were given regularly every 2·5 sec; Fig. 2b gives the results of a similar experiment in which the signals were given approximately every 5 sec and were triggered by the insect. Both methods give very similar results. In the first experiment the chirp rate was 55–60 per min and in the second 70–80 per min. In Fig. 2, the period of the interrupted cycle (y) is plotted against the chirp-to-signal period (x). The values of x and y were measured to the nearest 0·1 sec, and for each value of x the mean-to-minimum range of y is given; the chirp period has a positively skewed distribution, usually with a well-defined minimum. For comparison, the mean and minimum of the pre-signal chirp period (y₀) are indicated by horizontal lines. When the signal synchronizes with the chirp (– 0·05 < x ⪕ 0·05 sec) the interrupted period (y) is considered to begin with that chirp (inhibitory reaction time = 50 msec approx). When the ‘synchronous’ signal is triggered, the value of x is always greater than o; therefore in Fig. 2 b the values of y are plotted to the right of x = 0.

In Fig. 2 a the signal-to-chirp period (y—x) has a minimum value which decreases from 0·5 to 0·4 sec as x increases. In Fig. 2b, y — x has a minimum value of 0·4 sec, and the mean value of y has already begun to increase when x = 0·2. In Fig. 2,a, y is significantly greater than y₀ when x ⪖ 0·4 sec, and in Fig. 2 b, when x ⪖ 0·3 sec (P < 0·01, Wilcoxon matched-pairs signed-ranks test, Siegel, 1956).

This section summarizes the results of 39 experiments with seven isolated males. In each experiment a series of signals was given at the same phase of the cycle. The insect was allowed to chirp freely for at least 10 sec after each signal before the next one could be triggered. An average of 14 signals was given in each experiment and the insect was allowed to chirp freely for several minutes before and after each experiment.

During the first 50 msec after the beginning of a chirp (Fig. 3) the signal does not appear to affect the mean and median period of cycles y, y₂ and y₃. Nevertheless, the mode of the distribution of y is sharper than that of y₀, y₂ or y₃, suggesting that the period is more constant in the cycles in which the signals are given.

As the signals are given progressively later in the cycle they begin to affect period y, as shown in Fig. 2. In this second series of experiments an increase in period y was detected when the chirp-to-signal period was as little as 0·2 of the normal cycle (x/y₀ = 0·2). When the insect is chirping regularly, the period of subsequent cycles appears to be unaffected until x/y₀ ⪖ 0·4. Fig. 4 shows the effect of signals given late in the chirp cycle. As expected, y is significantly greater than y₀ (Wilcoxon test, P < 0·01); the mean, median and modal values of y are 0·4 – 0·5 sec greater than the corresponding values of y₀. The period of the next cycle (y₂) appears to be reduced by about 0·1 sec (Wilcoxon test, P < 0·05); this decrease in period lasts only for about one cycle (y₈ and y₀ are very similar). Thus the main effect of the signal is to delay and reset the cycle.

In an insect chirping irregularly, at a low rate, the first signals are often followed by a sudden increase in chirp rate and regularity (Fig. 5). In this case, the timing of the signal in the cycle does not appear to be particularly important. In the experiment illustrated in Fig. 6, the insect had been chirping fairly regularly with a chirp period of 1·1 – 1·5 sec, but had become very irregular in the previous half minute. After 6 – 7 signals the rate stabilized, but declined again after signal number 10. Fig. 7 shows the effect of a series of 16’synchronous’ signals (x = 0·05 sec) on the chirp pattern of another male. Previously, chirping had been irregular, the mean chirp period being approximately 2·5 sec. After five signals the chirp period became very regular and remained so for the rest of the test, becoming irregular again about 10 sec after the final signal.

When a male alternates with the signal the chirp rate is depressed, but during a prolonged period of alternation the rate gradually increases until it often equals or exceeds the previous uninterrupted rate. When the signals are stopped there is a ‘rebound’ increase in chirp rate. Fig. 9 combines the results of three tests; all were carried out with one male within a period of approximately 10 min. Similar results were obtained with other males.

The main effect of the artificial signal upon stridulation is to increase the period (y) of the cycle in which it is given. This is similar to the effect of a chirp by another male (Jones 1966a). When this happens, the cycle is reset and takes its phase from the delayed chirp. When the signal is synchronous with a chirp it does not affect the phase of the cycle, but may possibly make the cycle more regular. As the signal is given progressively later in the cycle, the period of the interrupted cycle (y) increases, but finally the cycle is so late (x ⪖ y₀) that it fails to inhibit the next chirp. Also, as the signal is given later in the cycle, the signal-to-chirp period (y — x) decreases to a minimum value, which possibly represents the integration time necessary for the production of a chirp, after the mechanism has been inhibited. In these experiments, and in alternation between singing males (Jones, 1966 a), this minimum integration time appears to depend on the basic chirp rate (or 1/y₀) and thus, presumably, on the excitation of the chirp-rhythm generator. Shaw (1968) has reported similar effects of natural and imitation chirps on the stridulation of P. camellifolia.

Previous experiments (Jones, 1966b) have shown that 10-msec signals have an inhibitory effect on the chirp cycle of Ph. griseoaptera which is almost identical to that of the 100-msec signals. As the signal length is increased to 1 sec or more, however, the signal-to-chirp interval decreases until eventually the insect chirps during the signal. Even then the chirp rate is normally lower during the signal than in the periods of silence between signals. The inhibitory effect of the signal may be considered as having phasic and tonic components. The initial effect stops and resets the chirprhythm generator; the inhibition may then decline towards the tonic level and the cycle is able to restart but with a much longer period. The reduction in signal-to-chirp interval may mean that the cycle has already progressed some way towards the production of another chirp before it is released from inhibition by the ending of the signal.

The signals may also affect subsequent cycles of the chirp-rhythm generator. In this series of experiments the after-effects were mainly excitatory, but Jones (1966b) has shown that such signals can also have inhibitory (or depressant) after-effects. If the insect is chirping regularly at a high rate, the excitatory after-effect is very small and can only be observed if the signal is given in the second half of the chirp cycle (x/y₀ ⪖ 0·4). This effect, a decrease in the chirp period, appears to last for only 1 – 2 cycles. This is similar to the observation by Heiligenberg (1969) that in A. domesticas about 99% of the excitatory effect of a stimulus chirp decays with a half-life of approximately 2 sec (phasic effect). Heiligenberg has also suggested that there is a smaller, tonic excitatory effect which decays with a half life of approximately 140 sec. In Ph. griseoaptera the excitatory effects of successive stimuli may summate to give slow, longer lasting increases in chirp rate. During alternation with a natural or artificial partner, the excitatory effects appear to be masked by the repeated inhibitory effects of the signals. Nevertheless, there is a gradual recovery of chirp rate. The total effect is only apparent when the insect is allowed to ‘escape’ from the alternation when the partner stops (Figs. 8 and 9).

The signals may have a much greater excitatory effect when the insect is chirping irregularly at a low rate. The chirp rate is increased, and the cycle becomes more regular (Figs. 5 and 6). My observations in the field and on caged males indicate that alternation with another singing male has a similar effect on a male which has been chirping in isolation at a low rate. It seems probable that, in this case, the excitation of the chirp-rhythm generator is very low before it is stimulated by the natural of artificial signals. The artificial signals appear to cause an increase in chirp rate, even when they are approximately synchronous with other chirps (Figs. 5b and 7). In these experiments the chirp length was approximately 70 msec and the ‘synchronous’ (100 msec) signals started 20 – 50 msec after the beginning of the chirp. The results indicate that a male can respond to such signals and therefore must be able to detect them in spite of the proximity of its own chirp. Further evidence for this is that repeated ‘synchronous’ signals appear to make the period of the interrupted cycle more regular (Fig. 3).

Although this paper is entirely concerned with the effects of artificial signals on stridulation in the male bush cricket, one cannot neglect the possible influence of the insect ‘s own song on its auditory system and thus on the control of sound production. No one has yet discovered what happens to the auditory system during the insect ‘s own chirp, and furthermore, we do not know how the auditory system is connected to the chirp-rhythm generator in the thorax. After their hearing organs have been removed, crickets and bush crickets chirp in an apparently normal manner (Fulton, 1928; Huber, 1963; Shaw, 1968; Jones & Dambach, 1973), although possible changes in syllable and chirp rate have not been investigated in detail. One would expect the intact tympanal organs to be grossly overloaded during the insect ‘s own song unless there is some mechanism for reducing the sound input or the sensitivity of the receptors. Nocke (1971) has shown that in G. campestris the stridulating front wings appear to function as a dipole sound source with the maximum output towards the posterior end of the insect. Thus it is possible that the hearing organs lie near a position of minimum intensity during the insect ‘s own song. In bush crickets the first pair of thoracic spiracles is very large, and the spiracles connect directly with the tympanal organs in the prothoracic legs. Lewis (1974) considers that the spiracles play an important part in controlling the auditory input; he found that in Homoro-coryphus nitidulus vicinus blocking the spiracles raises the threshold of the auditory nerve response by about 20 dB, while blocking the tympanal slits has little effect. Lewis ‘s measurements with a microphone probe indicate that the large thoracic spiracles are situated at points where the intensity of emitted sound is at a minimum. If, in addition, there is some mechanism for closing the spiracle or the associated air passage during the insect ‘s own song, this could serve as a mechanism for the control of sensitivity. It is also possible that such mechanisms could be reinforced by central inhibition as in bats (Suga & Schlegel, 1972) or by the efferent control of the sensory system as in the lateral line system of Xenopus (Russell, 1971).

Although auditory feedback seems to be unnecessary for normal stridulation, it is possible that it may reinforce the existing generator cycle. This would account for the apparent effect of ‘synchronous’ signals in making the cycle more regular, the signals adding to or prolonging the existing auditory feedback.

This research was carried out during the tenure of a Sir Henry Wellcome Travelling Fellowship from the Medical Research Council and was supported by the Deutsche Forschungsgemeinschaft. I thank Professor F. Huber for providing facilities and for his help and hospitality, and Professor Huber and Drs T. Collett and M. F. Land for their critical reading of the manuscript.