Recognition of Courtship Song in the Field Cricket, Teleogryllus Oceanicus

Home , Acheta, Teleogryllus, Teleogryllus commodus, Teleogryllus oceanicus

Anim. Behav., 1996, 51, 353–366

Recognition of courtship song in the ﬁeld cricket, Teleogryllus oceanicus

ROHINI BALAKRISHNAN & GERALD S. POLLACK Department of Biology, McGill University

(Received 4 November 1994; initial acceptance 9 February 1995; ﬁnal acceptance 9 June 1995; MS. number: 7146)

Abstract. The courtship song of the cricket, Teleogryllus oceanicus plays an important role in inducing the female to mount the male, which is necessary for mating. The song consists of a short, amplitude-modulated chirp, followed by a long trill of constant intensity and high syllable rate. Using playback techniques, it was determined which physical parameters of courtship song are necessary and/or sufficient to evoke normal female mounting of muted, courting males. The higher harmonics of natural courtship song were neither necessary nor sufficient for the effectiveness of the song. The chirp component alone was sufficient to evoke normal levels of mounting, but the trill was only partially effective on its own. The conspicuous amplitude modulation of the chirp was not necessary to evoke normal responses. The results suggest that the high effectiveness of the chirp is due to its characteristic temporal pattern. As in other cricket species, the song repertoire of T. oceanicus also includes distinct calling and aggression songs, which contain chirps that are structurally similar to the courtship chirp. Both calling and aggression songs evoked normal mounting responses when played back in the context of courtship. 1996 The Association for the Study of Animal Behaviour

Crickets use acoustic signals to communicate and courtship songs are recognized share common with each other. These signals take the form of or similar features. stereotyped, repetitive songs produced by stridu- Studies of song recognition have focused on lating males. Many cricket species, including Tel- calling song. Female crickets recognize and pref- eogryllus oceanicus, have a repertoire that includes erentially orient towards the conspecific calling three structurally distinct songs termed calling, song, which they identify based on its fundamen- courtship and aggression song, respectively tal frequency and species-specific temporal param- (Alexander 1961). The three songs serve distinct eters (reviewed in Elsner & Popov 1978; Doherty functions: calling song serves to attract females & Hoy 1985). The prior identification of song from afar, courtship song induces the female to parameters required for behavioural effectiveness mount the male, and aggression song is produced has been invaluable in studies of the neuronal during aggressive encounters with other males mechanisms for calling song recognition (Alexander 1962). (Schildberger et al. 1989). Little is known about The evolutionary processes that have resulted in the mechanisms, either behavioural or neuronal, this repertoire of three songs are unknown, but it of courtship song recognition in any cricket is generally believed that song in crickets first species. evolved in the context of courtship (Alexander The courtship behaviour of T. oceanicus, first 1962). The use of song for long-range attraction of described in detail by Burk (1983), is very similar females and in interactions with other males prob- to that of other gryllid species (Alexander 1961; ably evolved later. It is therefore interesting to Loher & Dambach 1989; Adamo & Hoy 1994). know whether the mechanisms by which calling Briefly, courtship is initiated when the antennae of the male contact the body of the female. The male then vibrates his antennae, adopts a typical singing posture and begins to produce courtship song, Correspondence: G. Pollack: Department of Biology, McGill University, 1205 Dr Penfield Avenue, simultaneously stroking the female with his anten- Montreal, Quebec H3A 1B1, Canada (email: nae. He then turns away from her and presents his [email protected]). abdomen, stridulating all the while. The female

0003–3472/96/020353+14 $12.00/0 1996 The Association for the Study of Animal Behaviour 353 354 Animal Behaviour, 51, 2

0 (a) (b) –20

–40

–60

–80 Relative intensity (dB) 0 6183012 24 0122461830 Frequency (kHz)

Chirp Trill

200 ms

(c)

30 ms Figure 1. The courtship song of T. oceanicus. (c) The oscillogram shows the amplitude-modulated chirp and long, constant-intensity trill. A portion of the trill is shown on an expanded time scale below. (a) Power spectrum of a single (high-amplitude) chirp pulse. (b) Power spectrum of a single trill pulse. Both chirp and trill pulses have a dominant peak at the fundamental frequency of 4–5 kHz and higher harmonics extending up to at least 30 kHz. Points superimposed on the spectra are mean (&) values of the relative intensity levels and frequencies of the harmonics. Data are based on analysis of at least 67 chirp and trill pulses in recordings from seven males.

reacts by following the male, approaching from success in eliciting a mounting response from the behind and touching his abdomen with her anten- females. Mounting can be restored if courtship is nae and palpi, whereupon the male spreads his accompanied by playback of recorded courtship hind wings and flattens his abdomen, allowing her song. to mount him. Mounting of the male by the The courtship song of T. oceanicus (previously female is a prerequisite for copulation. described by Leroy 1966; Hutchings & Lewis The role of courtship song in T. oceanicus was 1984; Libersat et al. 1994) consists of a 7–10 pulse, first examined by Burk (1983). From observations amplitude-modulated chirp (Fig. 1), followed by a of males under semi-natural conditions, he con- long trill (2–4 s) of high pulse rate (33/s). Both cluded that male mating success was correlated chirp and trill are produced at a fundamental with the production of courtship song. Muting frequency of 4.5 kHz and contain higher harmon- and playback experiments in different cricket ics extending up to about 30 kHz. Libersat et al. species support this conclusion (von Hörmann (1994) demonstrated that the fundamental 4.5- Heck 1957; Crankshaw 1979; Libersat et al. 1994). kHz component of courtship song was both Males that have been muted by removing their necessary and sufficient to evoke normal female tegmina perform all the acts of courtship nor- mounting responses in T. oceanicus, and also mally, including ‘singing’ (as seen by move- showed that one of the higher frequency compo- ments of their wing stubs), yet they have little nents, 13.5 kHz, was not sufficient. In the present Balakrishnan & Pollack: Cricket courtship song 355 study, we confirmed and extended their findings Spectral and Temporal Pattern Analysis on the exclusive importance of the fundamental Power spectra were computed from digitized frequency. We also investigated the importance of songs using the Fast Fourier Transform. Tem- the temporal parameters of the song. The goal of poral pattern analysis was performed either by these studies was to provide insight into the manually measuring pulse durations and pulse mechanisms underlying recognition of courtship periods from digitized recordings of songs, or by song. routing tape-recorded songs through a custom- built circuit that converted each sound pulse to a METHODS rectangular voltage signal (details of the circuit can be obtained from the authors). The onsets and Crickets offsets of these rectangular pulses were detected Teleogryllus oceanicus males and females were and stored on computer, and used to calculate reared in the laboratory at 25–30)C in 66-litre pulse durations and periods. The calculated plastic containers on a diet of Purina cat chow spectral and temporal parameters are reported as (Adult Formula) and given water ad libitum. mean values&. Male and female nymphs were separated before the final moult. Adult males and virgin Acoustic Stimuli females were kept separately in plastic boxes (38#20#23 cm) on a 12:12 h light:dark cycle for We delivered digitized sound pulses (12-bit at least 2 weeks before use. Experiments were resolution, 100-kHz update rate), either recorded performed within 3–6 h of the onset of scoto- or synthesized, at computer-controlled intervals. phase. We used 2–3-week-old virgin females and The signals were routed through a calibrated 2–4-week-old males in all experiments. attenuator (either Hewlett Packard 350D or Grason Stadler 1292) before being amplified (Sanken S1-1050G) and broadcast through a Song Recording and Analysis loudspeaker (Realistic 40-1310). Acoustic signals We recorded courtship, calling and aggression were digitized and synthesized with a National songs of 2–4-week-old T. oceanicus males in the Instruments ATMIO16F5 multi-function board, laboratory. To record courtship song, we intro- resident in a 80486-based computer. Signal analy- duced a male and a female into a cylindrical sis and synthesis were done with programs written (14#13 cm) cotton gauze cage. Recordings of with LabWindows software. calling song were obtained in the same manner, but from isolated males. To record aggression The Behavioural Assay song, we placed two males (of which one was muted by cutting off his forewings) in a smaller Courtship assays were performed in a cylindri- cylindrical cage (6#2.5 cm). All recordings were cal, anechoic chamber (13#10 cm) made of min- carried out in an anechoic chamber, in the dark, at eral wool covered with cheesecloth. A ring of a temperature of 24–26)C. cellulose acetate was pinned around the rim and We made recordings using a Brüel and Kjaer projected about 1 cm into the chamber to prevent condenser microphone (type 4135) and measuring the crickets from escaping. Experiments were amplifier (type 2610). The microphone was placed performed in the dark, at 24–26)C. Illumination at a height of 5–6 cm close to the wall of the cage was provided by a 60-W red light bulb suspended (because the male was free to move and turn in about 25 cm above and to one side of the any direction, the reported intensity levels are chamber. likely to be underestimates of the actual sound We selected courting males by pairing candi- intensity experienced by the female). The output dates with receptive females. Typically, the of the measuring amplifier was digitized (12-bit pair attempted repeatedly to mate, but was not resolution, 100-kHz sampling rate) and stored for permitted to do so. Males that showed courtship later analysis. Some songs were recorded on mag- behaviour were muted by cutting off both fore- netic tape (upper frequency limit: 20 kHz); these wings proximal to the file. They were allowed to were used only for analysis of temporal pattern. recover for 10–15 min and then, to ensure a high 356 Animal Behaviour, 51, 2 level of motivation, they were ‘pre-treated’ by differences in female response due to inter-male being given access to a female. After 5–10 min of variability because the same males were used in vigorous courtship had been observed, we used both control (playback of normal courtship song the male for experimental trials. We introduced or no playback) and test (playback of the test the male into the chamber and, after he became song) trials in a given experiment. We used a calm, introduced the female, as far away from the chi-squared test of independence with a table-wide male as possible. Typically, the female was im- significance level of 0.05 for each set of compari- mobile for several seconds after being introduced sons. The alpha errors were adjusted for multiple into the arena, but quickly recovered and started comparisons (within each figure in the Results) exploring the chamber. Occasionally, the male using the sequential Bonferroni procedure (Rice approached the female before she had recovered. 1989). We discarded these trials, as well as those in which either cricket ran around in the chamber without stopping. Males began to sing and court as soon Song Patterns as they encountered the female. In playback The structures of the test songs were derived experiments, each time the male attempted to sing from the measured features of natural T. a phrase of courtship song (as indicated by move- oceanicus songs. Unless stated otherwise, all song ments of the stumps of the forewings) a touch patterns that lacked amplitude modulation were of the computer keyboard by the experimenter played back at 90 dB SPL (peak intensity). For commanded the computer to play the test song song patterns with amplitude modulation, ampli- through a loudspeaker mounted directly above tude varied from 87 to 100 dB SPL, to mimic the the chamber at a height of 46 cm. The time from natural range of modulation. Songs consisting of the start of courtship ‘singing’ by the male to short chirps alone had a duration of 0.65 s; the mounting by the female was recorded. We termi- duration of all other song patterns was 3.7 s unless nated assays at 300 s in the absence of a mounting indicated otherwise. response. Two different types of ambiguous responses sometimes occurred. (1) Females occasionally mounted a male only partially, e.g. by Normal courtship song placing their forelegs on his abdomen without The normal courtship song stimulus consisted proceeding further. (2) Females sometimes repeat- of a nine-pulse chirp produced using the recorded edly, and rapidly, mounted and dismounted a pulses of a single chirp from a single male, whose male. We discarded these trials (approximately spectral and temporal features had values close to 5–7% of the total). We used each female for only the average (Figs 1, 7). The amplitude of each one trial. We used each male for eight to 10 trials, pulse was adjusted to match the average values which always included the controls (playback of shown in Fig. 7. The trill section of the song normal courtship song and absence of playback) consisted of a single representative trill pulse from as well as one or, usually, two different test the same male (peak intensity 90 dB SPL, dur- songs. ation 20 ms), iterated 100 times at periods of We used a sample size of 30 females for each 30.5 ms. Modified versions of courtship song were song pattern tested. In addition, we included the based on the foregoing parameters. Detailed two controls in each day’s experiments. In each explanations of their structures are presented in case, we calculated the percentage of females that the Results. mounted the male within 60 s from the start of courtship singing (the mounting frequency). There were no significant differences in the mounting Calling song (CALL) frequencies to the controls in 10 different experiments (no playback: ÷2=1.353, N=235, P=0.852; We constructed calling song using recorded normal song: ÷2=6.763, N=336, P=0.562). The pulses of natural calling song. We used three pulse mounting frequency for each test pattern, how- types (fundamental frequency=4.7 kHz): (1) a ever, was compared only to the controls in the chirp pulse of 39 ms duration (modal value, same experiment (N=30 for both control and N=685 chirp pulses), (2) a pulse of 25 ms duration experimental pattern). This procedure controls for (modal value=26 ms, N=762 trill pulses) for the Balakrishnan & Pollack: Cricket courtship song 357

first pulse of the trill doublet and (3) a pulse of mean duration of 34.0&6.1 ms (N=406 from 30 ms duration (modal value, N=762 trill pulses) seven males) and the mean pulse period (period for the second pulse of the trill doublet. A phrase from the onset of one pulse to onset of the consisted of five iterations of the chirp pulse (at following pulse) is 73.8&6.1 ms (N=358 from mean pulse period=63 ms, N=685) followed by seven males). The trill consists of one to three seven iterations (mean value, N=116) of the trill trains of pulses produced at a mean intensity of doublets. The intra-trill period was 35.5 ms (mean 91.5&1.3 dB SPL (N=30 from nine males). The value, N=762) and the inter-trill period was trill has a mean pulse duration of 17.9&2.8 ms 105.5 ms (mean value, N=645). The chirp–trill (N=18 600 from seven males) and the mean pulse period was 98 ms (mean value, N=110 phrases). period is 29.0&4.8 ms (N=18 771 from seven All the values reported here are based on data males). from the calling songs of seven males. The dur- Spectral analysis of chirp and trill pulses ation of the calling song was 1.23 s. To keep the revealed that both have a dominant, fundamental stimulus duration the same as that of the court- frequency of 4–5 kHz (Fig. 1), and higher har- ship song (3.7 s) during playback experiments, monics extending up to at least 30 kHz. There are the calling song stimulus was iterated three minor spectral differences between the different times. Calling song was played back at an in- pulse types of the song (Fig. 1). Although these tensity matched to that of courtship song, i.e. spectral differences are statistically significant with the trill at 90 dB SPL and the chirp at 92 dB (analysis not shown) they are small, on the order SPL. of 5–10 dB. Their biological relevance, if any, is unclear.

Aggression song (AGG) We used recorded pulses of a single, represen- Song is Necessary for Successful Courtship tative phrase of aggression song (fundamental frequency=4.4 kHz). To prevent confounding the Males that were rendered mute by cutting off results by the slight differences between the inten- their forewings showed normal courtship behav- sity profiles of courtship and aggression songs iour, including ‘singing’, which was evident from (Fig. 7), we played back the nine pulses of natural movements of the wing stubs. Compared with aggression song at the same intensity levels as intact males, however, they had little success in the courtship chirp (87–100 dB SPL). The pulse eliciting a mounting response (Fig. 2a, b; Mann– periods used were the calculated mean values for Whitney U-test: U=3448.0, P<0.001). pulses 1–9 in natural aggression song (Fig. 7). The When the singing movements of courting mute duration of the song was 0.68 s. males were accompanied by playback of normal courtship song, response frequency was enhanced (Fig. 2b, c; U=28421.0, P<0.001). The latencies of responses to mute males accompanied by RESULTS playback (median=27 s, exclusive of those cases where females failed to mount within 300 s) The Courtship Song were not significantly different from those evoked A typical phrase of courtship song consists of by intact males (median=20 s, U=6840.5, two parts: a short, amplitude-modulated chirp P=0.078). (mean& duration=621.5&72 ms, N=12 from Eighty-nine per cent of the females mounted 12 males), followed by a long trill (1.8–4.5 s, intact courting males within 60 s of the start mean& duration=2.7&0.96 s, N=13 from 13 of courtship singing (Fig. 2a). We therefore males) of almost constant intensity (Fig. 1). Both measured the proportion of females mounting chirp and trill consist of a series of discrete sound within this period (the mounting frequency) as an pulses or syllables. The chirp consists of 7–11 indicator of the effectiveness of courtship song. pulses (8.74&1.2, N=47 from 10 males). There is The mounting frequency in response to mute a conspicuous amplitude modulation, with a mean males in the absence of playback (16.7%) was difference in intensity between the first and last significantly different from that to intact, sing- pulses of the chirp of 13 dB. Chirp pulses have a ing males (89%; Fig. 2, ÷2=86.166, P<0.001). 358 Animal Behaviour, 51, 2

Playback of courtship song restored the mounting (a) N = 36 frequency to nearly normal levels (Fig. 2; 78% versus 89%, ÷2=2.337, P=0.126). 0.20

Higher Harmonics are neither Necessary nor 0.15 Sufficient for Recognition To investigate whether the higher harmonics are necessary for recognition of courtship song, 0.05 we constructed two songs consisting of the fundamental frequency alone. One of these (FILT: Fig. 3) consisted of recorded chirp and trill pulses from which the higher harmonics had been filtered (Butterworth bandpass, 4.0–5.0 kHz, order 10). 0.6 (b) The other (SYN: Fig. 3) was composed of syn- N = 235 thetic 4.5-kHz pulses: each chirp pulse had a rise 0.5 time equal to 55% and a fall time equal to 42% of its duration (mean value, N=35 pulses from six 0.4 males). Trill pulses (duration=19 ms) had rise and fall times of 11 and 7 ms, respectively (mean 0.3 values, N=39 pulses from six males). The temporal and amplitude-modulation features of both 0.2 songs were identical to the control, which was a phrase of normal courtship song (see Methods). Response frequency 0.1 All three songs were played back at 90 dB SPL (trill). In playback experiments, the songs that lacked higher harmonics elicited normal levels of mounting (Fig. 3; ÷2=0.131, P=0.718 for FILT; (c) ÷2=1.364, P=0.243 for SYN). Thus, the higher N = 336 harmonics are not necessary for recognition of 0.20 courtship song. To determine whether the higher harmonics 0.15 were sufficient, we constructed a synthetic song that consisted of the higher harmonics alone (HAR: Fig. 3). The spectral features of the syn- 0.10 thetic chirp and trill pulses reflected those of natural song (Fig. 1), except for the absence of the 0.05 fundamental. The temporal features of HAR were similar to those of natural courtship song. Most of the amplitude differences in the natural song, both 0 60120 180 240 300 within the chirp and between the chirp and trill, Time (s) are due to the fundamental component of the signal. When this component is removed by filter- Figure 2. Mounting responses to male courtship with ing, all of the sound pulses are nearly equal in and without song. Mounting response is the time elapsed amplitude, with peak intensities of 77 dB SPL. from the start of courtship singing by the male to HAR was therefore played back at this intensity. mounting by the female. (a) Response to normal, courting males with intact wings. (b) Response to courtship of This song was ineffective; response to the harmon- muted males in the absence of playback. (c) Response to ics alone was not significantly different from that courtship of muted males when accompanied by play- to mute males in the absence of playback back of normal courtship song at 90 dB SPL. The last (SILENT; Fig. 3; ÷2=0.303, P=0.582). Thus, the bar in (b) and (c) represents the frequency of females higher harmonics are not sufficient for recognition that did not mount within 300 s. of courtship song. Balakrishnan & Pollack: Cricket courtship song 359

100 No playback (SILENT) Normal song Test song Normal song Filtered song (FILT) Silent (4.5 kHz + harmonics) (4.5 kHz) 0 0 –20 80 –20 –40 –40 –60 –60 –80 –100 60 –80

0 6 12 18 24 30 0 6 12 18 24 30

Response (%) 40 Synthetic song (SYN) Harmonics alone (HAR) (4.5 kHz) alone 0 0 –20 –20 Relative intensity (dB) –40 –40 20 –60 –60 –80 –80 –100 –100 –120

0 0 6 12 18 24 30 0 6 12 18 24 30 FILT SYN HAR Figure 3. The role of the higher harmonics in courtship song recognition. This and subsequent graphs show the percentage of females that mounted mute, courting males within 60 s from the start of courtship singing. Each pair of open and solid bars shows the percentage mounting to normal song and test pattern, respectively, in a given experiment. Circles above the bars indicate response frequencies signiﬁcantly diﬀerent from those to playback of normal song in the same experiment. N=30 in each case.

The Temporal Pattern of the Chirp is Sufficient to poral pattern and not its amplitude modulation. Evoke Normal Responses We tested this further by presenting a song The effectiveness of the chirp and trill compo- (CHRP-AM) that retained the temporal features nents was tested by playing them back indepen- of the natural chirp but lacked amplitude modu- dently. The long natural trill, when presented lation (Fig. 5). This non-amplitude-modulated alone (TRILL), was not sufficient to restore chirp (produced by nine iterations of a single, the mounting response to normal levels (Fig. 4; 4.5 kHz, 35-ms chirp pulse at pulse periods of ÷2=5.554, P=0.036). A song consisting of the 73.8 ms) was also sufficient to evoke normal natural chirp alone (CHRP) however, was suf- levels of mounting (Fig. 5; ÷2=2.857, P=0.091). ficient to evoke mounting responses that were To confirm that the amplitude modulation of the not significantly different from normal (Fig. 4; chirp was not necessary in the context of the ÷2=2.857, P=0.091). normal, two-part song, we replaced the amplitude- The relatively high effectiveness of the chirp modulated chirp of the normal song with a single, compared to the trill component could be due 4.5-kHz chirp pulse (duration=35 ms), iterated either to the characteristic temporal pattern of the nine times (pulse period=73.8 ms) at constant chirp or to its conspicuous amplitude modulation. intensity (90 dB; i.e. the same intensity as the trill). To test this, we constructed a song (CONT This constant-intensity song (NORM-AM) was as CHRP: Fig. 5) that differed from the natural trill effective as the normal song (Fig. 5; ÷2=0.0, only in temporal pattern. This song was produced P=1.0), confirming that amplitude modulation is by 50 iterations of a single, 4.5-kHz chirp pulse not necessary for recognition of courtship song. (duration=35 ms) at pulse periods of 73.8 ms. The The relative ineffectiveness of the trill afforded response to this stimulus (CONT CHRP) did not us an opportunity to determine whether the differ significantly from that to the normal song amplitude modulation that occurs during the (Fig. 5; ÷2=0.317, P=0.573), suggesting that the chirp, while not necessary, might nevertheless play high effectiveness of the chirp is due to its tem- a role in recognition. We imposed amplitude 360 Animal Behaviour, 51, 2

100 Normal song Test song Normal song

60 Chirp alone (CHRP)

40 Response (%)

20 Trill alone (TRILL)

0 CHRP TRILL Figure 4. Eﬀectiveness of the chirp and trill components in evoking the mounting response. N=30 in each case.

100 Normal song Test Normal song song

Normal song without 60 amplitude modulation (NORM – AM)

40 Response (%) Continuous chirp (CONT CHRP)

0 Chirp without amplitude NORM – CONT CHRP – modulation (CHRP – AM) AM CHRP AM

Figure 5. Relation between the conspicuous amplitude modulation and the eﬀectiveness of the chirp. CONT CHRP and CHRP-AM: Songs that retained the temporal pattern of the chirp but lacked amplitude modulation; NORM-AM: a two-part song that lacked amplitude modulation. N=30 in each case.

modulation on the temporal pattern of the trill in AMTR2, the modulation was applied over a time two ways (Fig. 6). In the song AMTR1, the period similar to the duration of the natural chirp modulation occurred over the ﬁrst nine pulses of (610 ms), by changing amplitude every second the trill, as it normally does in the chirp. In pulse for the ﬁrst 18 pulses. In both cases, the Balakrishnan & Pollack: Cricket courtship song 361

100 Normal song Normal song Test song 80

60 Amplitude modulation over first 9 pulses of trill (AMTR1)

40 Response (%)

20 Amplitude modulation of trill over duration of chirp (AMTR2)

0 AMTR1 AMTR2 Figure 6. Relation between amplitude modulation and the effectiveness of the trill. A 13-dB amplitude modulation (AM) was imposed on the temporal pattern of the natural trill (played back at 90 dB SPL). In the first case (AMTR1), the AM was imposed over the first nine pulses of the trill (same number as the natural chirp). In the second case (AMTR2), the AM was applied over 610 ms (to mimic the duration of the natural chirp). N=30 in each case.

range of amplitude modulation was identical to consists of a chirp (Fig. 7) of seven to 22 pulses that of natural chirp (13 dB) and was achieved by (11.2&3.0, N=93 from 12 males). adjusting the peak-to-peak amplitudes of nine trill The chirps of calling, courtship and aggression pulses to the mean intensity levels of successive song are similar in their spectral features (Leroy pulses in natural chirps. AMTR1 restored the 1966; Loher & Rence 1978; Hutchings & Lewis response to a level not significantly different from 1984). All have a dominant, fundamental fre- normal (Fig. 6; ÷2=2.7, P=0.1), showing that quency of 4–5 kHz, with higher harmonics extend- amplitude modulation can enhance the effective- ing up to 30 kHz (Fig. 7). The second harmonic is ness of a temporal pattern that is less attractive at least 20 dB less intense than the fundamental in than that of the chirp. AMTR2, however, did not all three chirps, and the higher harmonics are even restore the response to normal levels (Fig. 6; less intense. Detailed statistical analysis (not ÷2=6.787, P=0.018). In fact, the response to shown) revealed some minor (5–10 dB) but statis- AMTR2 was almost identical to that evoked by tically significant differences between the mean the trill without amplitude modulation (40% relative intensity levels of some of the harmonics. versus 43%: see Fig. 4). The aggression song is very similar to the courtship chirp in pulse number (mode=9 in both cases), pulse durations and pulse periods (Fig. 7) Calling and Aggression Songs can also Elicit and in the presence of conspicuous amplitude Normal Mounting Responses modulation. The calling chirp has fewer pulses, Like the courtship song, the calling song of lacks amplitude modulation and has a shorter T. oceanicus (Leroy 1966; Bentley & Hoy 1972; mean pulse period, but is similar in terms of pulse Hill et al. 1972) also consists of two parts: a short duration (Fig. 7). chirp of four to eight pulses (X&=5.9&1.0, On the basis of the structural similarities of the N=116 from seven males) followed by a complex three chirps described above, we predicted that trill that consists of six to 10 pulse doublets aggression and calling song should also elicit (6.9&1.1, N=116 from seven males). The aggres- responses in the context of courtship. To test this sion song (Leroy 1966; Hutchings & Lewis 1984) prediction, we performed playback experiments 362 Animal Behaviour, 51, 2

(a) Courtship Calling Aggression

(b) 0

–20

–40

–60 Intensity (dB) –80 0 61812 24 30 0122461830 01224618 30 Frequency (dB) 110 50 100 (c) (d) (e) 90 40 100 80 30 Courtship 70 90 Aggression Period (ms) Intensity (dB) Duration (ms) 20 60 Calling 80 10 50 1 23456789 1 23456789 1 23456789 Chirp pulse number Chirp pulse number Pulse period number Figure 7. Structure of courtship, calling and aggression chirps. (a) Oscillograms of representative chirps of courtship, calling and aggression songs. (b) Power spectra of single courtship, calling and aggression chirp pulses. Data points in the spectra represent means (&s) of the relative intensity levels and frequencies of the harmonics. Data are based on analysis of 67 (courtship), 21 (calling) and 71 (aggression) chirp pulses from seven males in each case. (c) Graph shows the mean (+) intensity level (dB SPL) of each successive pulse of courtship (N=9), calling (Nd17) and aggression (N=10) chirps. (d) Graph shows the mean (+) duration of successive pulses in courtship (Nd24), calling (Nd77 for pulses 1–6, 29 for pulse 7) and aggression (Nd17) chirps. (e) Graph shows mean (+) pulse period for successive pulses in courtship (Nd24), calling (Nd79 for periods 1–5, 29 for period 6) and aggression (Nd17) chirps. Recordings were obtained from seven males. with recorded calling and aggression songs. Nor- Burk (1983) observed that T. oceanicus females mal mounting responses (Fig. 8) were evoked by would not mount males that did not produce both aggression (÷2=1.920, P=0.166) and calling courtship song. However, some studies on Acheta songs (÷2=0.098, P=0.754). (Ghouri & McFarlane 1957) and T. commodus (Loher & Rence 1978) did not reveal a significant DISCUSSION decrease in male mating success as a result of muting males or deafening females. Differences in Role of Song in Courtship results could be due to differences in behavioural Our results agree with earlier work on field paradigms, including factors such as duration of crickets that demonstrated the importance of song the assay, arena size, age and prior exposure of as a courtship signal. Muted males had little females to courting males (Adamo & Hoy 1994). success in eliciting a mounting response from Muting experiments in the field have confirmed females (von Hormann-Heck 1957; Crankshaw the importance of song for male mating success in 1979; Adamo & Hoy 1994; Libersat et al. 1994). grasshoppers (Kriegbaum & von Helversen 1992) Balakrishnan & Pollack: Cricket courtship song 363

100 Courtship song Test song Courtship song

Calling song (CALL)

40 Response (%)

20 Aggression song (AGG)

0 CALL AGG Figure 8. Response to calling and aggression songs in the context of courtship. N=30 in each case. and sagebrush crickets (Snedden & Sakaluk 1992). when the male was not singing, so that their Acoustic signalling as a necessity for successful females were exposed to much greater levels of courtship is well documented for the hissing acoustic stimulation than in our case. Also, cockroach, Gromphadorhina portentosa (Nelson Libersat et al. used a different experimental design & Fraser 1980), the tree cricket, Oecanthus that involved repeated testing of individual pairs nigricornis (Bell 1980) and several species of of crickets. Drosophila (Ewing 1989; Bixler et al. 1992; We found that the chirp component of court- Liimatainen et al. 1992). In most cases, however, ship song was sufficient to evoke normal mounting muting did not completely abolish male mating responses, the trill being dispensable as an acoustic success, showing that song is not an absolute cue. Our results indicate that the mechanism for requirement for successful courtship. courtship song recognition in T. oceanicus is temporal pattern-dependent. Temporal pattern is the major cue for recognition in many signalling The Necessary Acoustic Parameters of systems; for example, calling song recognition in Courtship Song orthopterans (Elsner & Popov 1978; Ewing 1989) Our experiments indicate that the two necessary and recognition of the vibrational mating songs acoustic parameters for an effective courtship of planthoppers, Nilaparvata lugens (Claridge song are the fundamental frequency (4.5 kHz) and et al. 1984) and green lacewings, Chrysoperla the temporal pattern of the chirp. Our results plorabunda (Wells & Henry 1992). Temporal pat- agree with those of Libersat et al. (1994), who tern is also important for recognition of courtship demonstrated that the fundamental frequency of song in D. melanogaster (Bennett-Clark & Ewing T. oceanicus courtship song is both necessary and 1969; Kyriacou & Hall 1982), and is the major cue sufficient to evoke normal mounting responses. in the vibratory courtship signals of the wandering In contrast to their results, however, we found spider, Cupiennius salei (Schüch & Barth 1990). that the courtship trill presented on its own was Neurons that show selectivity for temporal pattern only partially effective and not sufficient to elicit exist in the central nervous systems of a variety of normal levels of mounting. The discrepancy in organisms (Rose 1986). Bandpass filter neurons our results could be due to differences in the that are tuned to the temporal pattern of calling paradigms used: their assays were of longer dur- song have been identified in the brain of the ation (5 min versus 1 min) and their stimuli were field cricket, Gryllus bimaculatus (Schildberger played continuously during this period, even 1984). 364 Animal Behaviour, 51, 2

The Role of Amplitude Modulation she is distant from the male, but by mounting the male if the two have come into contact. Similarly, The conspicuous amplitude modulation of the chirps help males to gain and maintain social chirp was not necessary for its function. When dominance or territory (Nolen et al. 1992) when amplitude modulation was imposed over the first produced during close-range interactions with nine pulses of the trill temporal pattern, however, other males, but evoke mounting responses if it enhanced the effectiveness of the trill so that the recipient is female. This dependence of the mounting was restored to nearly normal levels. response on factors extrinsic to the signal, such as This result implies that the amplitude modulation sex, context and additional sensory inputs, was present in courtship song can be evaluated by the foreseen more than 30 years ago by Alexander nervous system, but its effect is redundant when (1962). the temporal pattern is highly attractive. Our In T. oceanicus, then, we speculate that a single results with the second amplitude-modulated trill acoustic signal, the chirp, detected by a single pattern (AMTR2) indicate that the parameter that neuronal recognizer, has evolved diverse func- is used to evaluate amplitude modulation is the tions: long-range mate attraction during calling, number of pulses rather than the duration over enhancing the probability of female mounting which it occurs. during courtship, and determining male social There is evidence for the use of amplitude status. modulation for song recognition in some orthopterans; in the recognition of calling song in the cricket, Melanogryllus desertus (Elsner & Popov The Role of the Trill Components 1978), and the grasshopper, Chorthippus biguttulus Why do calling and courtship songs contain trill (von Helversen & von Helversen 1983). components if their chirps are sufficient to attract mates and elicit female mounting? One possibility A Single ‘Chirp Recognizer?’ is that the chirp and trill serve different functions: the different components of a complex signal may Calling and courtship song recognition in be directed to different receivers, as in the case T. oceanicus share striking similarities. Both songs of the frog, Eleutherodactylus coqui (Narins & are complex, consisting of chirp and trill compo- Capranica 1976). Pollack (1982) showed that, in nents. In both cases, the song parameters that are flight phonotaxis assays, T. oceanicus males sig- important to evoke normal responses (female nificantly preferred the temporal pattern of the phonotaxis to calling song or female mounting in calling trill to that of the chirp, raising the possi- response to courtship song) are the fundamental bility that the calling trill is involved in male–male frequency of 4–5 kHz and the temporal pattern of interactions. Evidence suggests that T. oceanicus the chirp (Moiseff et al. 1978; Pollack & Hoy males form aggregates in the wild (Cade 1981), 1981). In calling song, as in courtship song, the but the possible role of the calling trill in this chirp temporal pattern is sufficient to elicit normal behaviour has not been tested. responses (Pollack & Hoy 1981). The chirps of Although our experiments indicate that the calling, courtship and aggression songs are very courtship trill is dispensable as an acoustic cue, we similar in structure and all three songs were believe that it plays an important role during equally effective in eliciting mounting responses in courtship for two major reasons. (1) Males pro- the context of courtship. This similarity suggests gressively increase the length of their trills as the that a single neuronal filter (‘chirp recognizer’) courtship bout progresses (Libersat et al. 1994; may mediate the responses to calling, courtship personal observations), even though this is likely and aggressive chirps. to be energetically expensive (Prestwich & Walker If the above speculation is correct, then the 1981). (2) Burk (1983) observed that females did different behavioural responses to the three songs not mount males that produced only chirps. In may depend not primarily on their mediation by our experiments, the chirp alone was very effective different signal recognizers but rather on factors in eliciting mounting. We believe that the reason such as the sex of the recipient and on behavioural for the discrepancy in the two observations is that context. For example, a receptive female responds in our experiments with playback of the chirp to the chirp pattern by performing phonotaxis if alone, the muted male was still performing the Balakrishnan & Pollack: Cricket courtship song 365 movements associated with trilling. Stridulatory Sexual Competition in a Diverse Group of Insects (Ed. movements generate vibrations that could be by D. T. Gwynne & G. K. Morris), pp. 97–119. perceived by the female either through the sub- Boulder, Colorado: Westview Press. Cade, W. H. 1981. Field cricket spacing, and the strate or through direct body contact during phonotaxis of crickets and parasitoid flies to clumped courtship (Gogala 1985). We speculate that, and isolated cricket songs. Z. Tierpsychol., 55, 365– during courtship, two important signals are nec- 375. essary to evoke normal mounting: an acoustic Claridge, M. F., den Hollander, J. & Morgan, J. C. 1984. Specificity of acoustic signals and mate choice in component provided primarily by the courtship the brown planthopper Nilaparvata lugens. Entomol. chirp, and a non-acoustic component (perhaps exp. appl., 35, 221–226. vibration) provided largely by the trill. Crankshaw, O. S. 1979. Female choice in relation to The observed dispensability of the acoustic calling and courtship songs in Acheta domesticus. component of the trill in our experiments could Anim. Behav., 27, 1274–1275. Doherty, J. & Hoy, R. 1985. Communication in insects. also be ascribed to an age-dependent loss of III. The auditory behavior of crickets: some views of selectivity in T. oceanicus females. We used 2–3- genetic coupling, song recognition, and predator week-old females in all our experiments: younger detection. Q. Rev. Biol., 60, 457–472. females may be more selective (Walikonis et al. Elsner, N. & Popov, A. V. 1978. Neuroethology of acoustic communication. Adv. Insect Physiol., 13, 1991) and may require the acoustic component of 229–355. the trill for normal mounting. Ewing, A. W. 1989. Arthropod Bioacoustics; Neuro- biology and Behavior. Ithaca, New York: Cornell University Press. ACKNOWLEDGMENTS Ghouri, A. S. K. & McFarlane, J. E. 1957. Reproductive isolation in the house cricket (Orthoptera: Gryllidae). We thank two annonymous referees whose com- Psyche, 64, 30–36. Gogala, M. 1985. Vibrational communication in insects ments considerably improved the manuscript. (biophysical and behavioural aspects). In: Acoustic This work was supported by the Natural Sciences and Vibrational Communication in Insects (Ed. by K. and Engineering Research Council of Canada. Kalmring & N. Elsner), pp. 117–126. Berlin: Verlag Paul Parey. von Helversen, D. & von Helversen, O. 1983. Species recognition and acoustic localization in acridid grass- REFERENCES hoppers: a behavioral approach. In: Neuroethology and Behavioral Physiology (Ed. by F. Huber & H. Adamo, S. A. & Hoy, R. R. 1994. Mating behaviour of Markl), pp. 95–107. Berlin: Springer-Verlag. the field cricket Gryllus bimaculatus and its depend- Hill, K. G., Loftus-Hills, J. J. & Gartside, D. F. 1972. ence on social and environmental cues. Anim. Behav., Premating isolation between the Australian field 47, 857–868. crickets Teleogryllus commodus and T. oceanicus Alexander, R. D. 1961. Aggressiveness, territoriality, (Orthoptera: Gryllidae). Aust. J. Zool., 20, 153– and sexual behavior in field crickets (Orthoptera: 163. Gryllidae). Behaviour, 17, 130–223. von Hörmann-Heck, S. 1957. Untersuchungen über den Alexander, R. D. 1962. Evolutionary change in cricket Erbgang einiger Verhaltenwiesen bei Grillenbastarden acoustical communication. Evolution, 16, 443–467. (Gryllus campestris L.Gryllus bimaculatus DeGeer). Bell, P. D. 1980. Multimodal communication by the Z. Tierpsychol., 14, 137–183. black-horned tree cricket, Oecanthus nigricornis Hutchings, M. & Lewis, B. 1984. The role of two-tone (Walker) (Orthoptera: Gryllidae). Can. J. Zool., 58, suppression in song coding by ventral cord neurones 1861–1868. in the cricket Teleogryllus oceanicus (Le Guillou). Bennett-Clark, H. C. & Ewing, A. W. 1969. Pulse J. comp. Physiol. A, 154, 103–112. interval as a critical parameter in the courtship song Kriegbaum, H. & von Helversen, O. 1992. Influence of of Drosophila melanogaster. Anim. Behav., 17, 755– male songs on female mating behavior in the grass- 759. hopper Chorthippus biguttulus (Orthoptera: Acridi- Bentley, D. R. & Hoy, R. R. 1972. Genetic control of dae). Ethology, 91, 248–254. the neuronal network generating cricket (Teleogryllus Kyriacou, C. P. & Hall, J. C. 1982. The function of Gryllus) song patterns. Anim. Behav., 20, 478–492. courtship song rhythms in Drosophila. Anim. Behav., Bixler, A., Jenkins, J. B., Tompkins, L. & McRobert, 30, 794–801. S. P. 1992. Identification of acoustic stimuli that me- Leroy, Y. 1966. Signaux acoustiques, comportement diate sexual behavior in Drosophila busckii (Diptera: et systématique de quelque espèces de Gryllides- Drosophilidae). J. Insect Behav., 5, 469–478. (Orthoptères, Ensifères). Bull. biol. Fr. Belg., 100, Burk, T. 1983. Male aggression and female choice in a 1–134. field cricket (Teleogryllus oceanicus): the importance Libersat, F., Murray, J. A. & Hoy, R. R. 1994. Fre- of courtship song. In: Orthopteran Mating Systems: quency as a releaser in the courtship song of two 366 Animal Behaviour, 51, 2

crickets, Gryllus bimaculatus (deGeer) and Teleogryl- Prestwich, K. N. & Walker, T. J. 1981. Energetics of lus oceanicus: a neuroethological analysis. J. comp. singing in crickets: effect of temperature in three Physiol. A, 174, 485–494. trilling species (Orthoptera: Gryllidae). J. comp. Liimatainen, J., Hoikkala, A., Aspi, J. & Welbergen, P. Physiol., 143, 199–212. 1992. Courtship in Drosophila montana: the effects of Rice, W. R. 1989. Analyzing tables of statistical tests. male auditory signals on the behaviour of flies. Anim. Evolution, 43, 223–225. Behav., 43, 35–48. Rose, G. 1986. A temporal-processing mechanism for all Loher, W. & Dambach, M. 1989. Reproductive behav- species? Brain Behav. Evol., 28, 134–144. ior. In: Cricket Behavior and Neurobiology (Ed. by F. Schildberger, K. 1984. Temporal selectivity of identified Huber, T. E. Moore & W. Loher), pp. 43–82. Ithaca, auditory neurons in the cricket brain. J. comp. New York: Cornell University Press. Physiol., 155, 171–185. Loher, W. & Rence, B. 1978. The mating behavior of Schildberger, K., Huber, F. & Wohlers, D. W. 1989. Teleogryllus commodus (Walker) and its central and Central auditory pathway: neuronal correlates of peripheral control. Z. Tierpsychol., 46, 225–259. phonotactic behavior. In: Cricket Behavior and Moise , A., Pollack, G. S. & Hoy, R. R. 1978. Steering ff Neurobiology (Ed. by F. Huber, T. E. Moore & W. responses of flying crickets to sound and ultrasound: Loher), pp. 423–459. Ithaca, New York: Cornell mate attraction and predator avoidance. Proc. natn. University Press. Acad. Sci. U.S.A., 75, 4052–4056. Schüch, W. & Barth, F. G. 1990. Vibratory communi- Narins, P. M. & Capranica, R. R. 1976. Sexual differences in the auditory system of the tree frog Eleuthero- cation in a spider: female responses to synthetic male J. comp. Physiol. A dactylus coqui. Science, 192, 378–380. vibrations. , 166, 817–826. Nelson, M. C. & Fraser, J. 1980. Sound production in Snedden, W. A. & Sakaluk, S. K. 1992. Acoustic sig- the cockroach, Gromphadorhina portentosa: evidence nalling and its relation to male mating success in for communication by hissing. Behav. Ecol. Sociobiol., sagebrush crickets. Anim. Behav., 44, 633–639. 6, 305–314. Walikonis, R., Schoun, D., Zacharias, D., Henley, J., Nolen, T. C., Lam, C., Wong, J., Luayon, L. & Luayon, Coburn, P. & Stout, J. 1991. Attractiveness of the J. 1992. High frequency components in the rivalry male Acheta domesticus calling song to females. III. song of territorial male crickets: multiple functions for The relation of age-correlated changes in syllable aversion to ultrasound. Proc. Int. Congr. Neuroethol., period recognition and phonotactic threshold to 3. juvenile hormone III biosynthesis. J. comp. Physiol. A, Pollack, G. S. 1982. Sexual differences in cricket calling 169, 751–764. song recognition. J. comp. Physiol., 146, 217–221. Wells, M. M. & Henry, C. S. 1992. Behavioural Pollack, G. S. & Hoy, R. R. 1981. Phonotaxis to responses of green lacewings (Neuroptera: Chrysopi- individual rhythmic components of a complex cricket dae: Chrysoperla) to synthetic mating songs. Anim. calling song. J. comp. Physiol., 144, 367–373. Behav., 44, 641–652.