A Paradoxical Problem in Insect Communication: Can Bush Crickets Discriminate Frequency?

The stridulatory apparatus of bush crickets is an elaborately constructed device with design features peculiar to each species. When activated, the resonating area produces a characteristic modal vibration -the so-called carrier frequency of the call. This is inbuilt and cannot be altered by the insect. Other features of the call -tooth strike rate, amplitude modulation, syllable length and syllable sequence -could be altered, since they are theoretically controlled by the insect’s nervous system. If the tooth strike rate is such that the wing resonator is synchronously excited then a long harmonic call can result, as in Ruspolia differens (Serville) (previously Homorocoryphus) (Bailey, 1970) and Mecopoda elongata (Linnaeus) (J. C. Hartley and R. O. Stephen, personal observation). Such calls present no problem in frequency determination. If, however, the tooth strike rate is low, each impulse may produce only a few oscillations which decay well before the next impulse, as shown by the examples in Fig. 1 and given for many other species (Table 5.1, Sales and Pye, 1974). The use of brief harmonic transients is therefore by no means a rare occurrence in the Tettigoniidae. If these brief transients are to have a communication value, is their frequency important or is it just the pulse pattern that identifies the call? If the frequency is important, how can it be determined?

Taking the call of Conocephalus dorsalis as an example (Fig. 1C), each tooth strike generates a transient with only two oscillations before decaying, leaving a long and variable gap before the next impulse. The frequency of these oscillations, estimated from the trace, is 44kHz. The calculated frequency uncertainty is ±11 kHz (equation 1). The power spectrum of a single tooth strike (Fig. 1D) also shows frequency uncertainties of 12 kHz at the -3dB level and 23 kHz at -5dB (upper and lower bars, respectively, in Fig. 1D). Thus, precise determination of the frequency of the transient is impossible, illustrating the type of acoustic paradox confronting the insect.

Since the harmonic transient is generated by the vibration of the highly specialised wing resonator, which is peculiar to each individual species, it must be assumed that the insect has a way of receiving the information contained within the transient. This information could become available if the ear contained an additional high Q resonator. Stephen and Bailey (1982) suggested that the air pockets contained by the external tympanal chambers (slit cavities) of the ears of Hemisaga could act as as pair of coupled Helmholtz resonators. A feature of Helmholtz resonators is their high value of Q. The effect of including such resonators into the ear would be to extend the overall duration of each pulse, thereby effectively increasing the number of cycles in each transient. This should significantly improve the frequency discrimination.

To test this, we constructed a Helmholtz resonator which could be attached to the front of a 0.5 in (12.5 mm) Bruel & Kjaer microphone (Fig. 3). The frequency response of the modified microphone (Fig. 4A) is now highly tuned with a centre frequency of 2.6 kHz. An Audax tweeter was then driven to produce a transient pulse at this frequency. The transient signal was initially recorded through the microphone without the resonator (Fig. 4B). The resonator was re-attached and the transient recorded again (Fig. 4C). The addition of the resonator has clearly enhanced both amplitude and duration of the transient. Changing just the frequency of the transient, but keeping the number of oscillations, amplitude and relative positions the same, produced results shown in Fig. 4D,E. At both higher and lower frequencies, not only is the perceived pulse shorter but it is also of much lower intensity. One problem resulting from these very brief transients is that rise time and envelope shape will both contribute substantially to the convoluted sound (the overall combined sound, see Randall, 1977) to which our system was tuned. Our relatively crudely constructed Helmholz resonator is unlikely to perform as well as those with several million years’ refinement behind them.

Many Tettigoniidae have developed slit chambers over their ears. There seems to be no generally accepted explanation for the function of these structures, although it has been suggested that in Hemisaga they could improve location of the species call and hence aid phonotaxis (Stephen and Bailey, 1982). We now suggest that they may also enable frequency determination of harmonic transients that would otherwise be too brief to resolve. This argument is particularly applicable to ephippigerines, which all appear to use extremely brief transients lasting no more than three or four oscillations, as exemplified by Steropleurus ortegai (Fig. 1A). That the rise time of the perceived transient will be slightly delayed matters not, as the relative time intervals will stay the same. Auditory sensitivity may also benefit from an improvement in the signal to noise ratio which will inevitably follow from an enhanced frequency sensitivity.

The Mecopodinae and many Phaneropterinae have exposed tympana with no development of slit chambers. Mecopoda elongata with its long harmonic call, as previously mentioned, needs no enhancement system. The phaneropterines Ancistrura nigrovittata (Heller, 1988), Poecilimon schmidti (Hartley and Stephen, 1989), Leptophyes punctatissima and L. laticauda (J. C. Hartley and R. O. Stephen, personal observations) produce calls in which each tooth impulse produces a brief tone burst consisting of at least six oscillations. Such pulses will not present the extreme frequency resolution problems that appear with the Ephippigerinae.

It might be argued that it is the tooth-strike frequency and not the resonant frequency of the wing that is important in species such as Conocephalus. However, as shown in Fig. 1C, this is clearly not regular enough to generate a distinctive frequency. In behavioural experiments on Requena, Bailey and Yeoh (1988) found that females were able to differentiate between 16 and 28 kHz, but were unable to discriminate between 28 and 36 kHz. This is rather better than the signal frequency uncertainty apparent in Conocephalus. Kalmring et al. (1990) report that transient pulses at the right frequency are most important components in many bush cricket communication systems. The tettigoniid ear is known to have a spatial arrangement of receptors associated with particular frequencies (reviewed by Rheinlaender and Romer, 1990).

We wish to thank Dr H. C. Bennet-Clark for useful comments on an earlier version of this report.