Moths are not silent, but whisper ultrasonic courtship songs

Most moths have tympanal ears sensitive to ultrasound. The tuning of moth hearing to bat calls (Fullard, 1998) as well as the deterioration of moth hearing in bat-free areas (Surlykke et al., 1998) indicates that ears probably evolved in response to selective pressures imposed by foraging bats (Hoy and Robert, 1996; Miller and Surlykke, 2001; Waters, 2003). The location and morphology of moth ears vary across moth superfamilies, indicating an independent development in each group after the divergence of moth superfamilies (Hoy and Robert, 1996; Greenfield, 2002; Yack, 2004).

Subsequent to the development of ears, a relatively small number of moth species developed sound producing organs, and used them either for defense against bats or rival males (Spangler, 1988; Conner, 1999; Greenfield, 2002; Waters, 2003). For instance, both sexes of tiger moths (Arctiidae) emit jamming or warning ultrasonic clicks in response to echolocation calls of bats (Surlykke and Miller, 1985; Barber and Conner, 2007; Ratcliffe and Nydam, 2008). In the lesser wax moth, Achroia grisella (Pyralidae), males produce loud ultrasonic clicks to attract female mates (Spangler, 1988). The ultrasounds used by these species are characterized by high sound pressure levels, with the peak equivalent sound pressure level (peSPL) at a distance of 1 cm ranging from 76 to 125 dB.

Production of ‘courtship’ sounds, the sounds produced after pair formation and before mating, has only rarely been reported in moths, e.g. in some arctiids (Conner, 1987; Krasnoff and Yager, 1988; Sanderford and Conner, 1990; Simmons and Conner, 1996; Sanderford et al., 1998) and pyralids (Spangler, 1985; Trematerra and Pavan, 1995). The acoustic communication between sexes has not been verified in most cases, but one effect of the male songs seems to be to prompt their mates to accept mating, in some cases in conjunction with chemical signals as in the pyralid, Galleria mellonella (Spangler, 1985). Given that many arctiid moths produce ultrasonic clicks to counteract bats (Miller and Surlykke, 2001), sexual communication via ultrasounds is likely to be a secondary use of the ability to produce ultrasound (Endler and Basolo, 1998).

We recently added the Asian corn borer moth, Ostrinia furnacalis (Crambidae), to the list of moths that produce courtship songs. O. furnacalis shows mating behavior typical of the majority of moths. First, males of O. furnacalis are attracted toward a sex pheromone-releasing female. After landing near the female, the male approaches her with his wings fanning (<2 cm) and subsequently attempts to mate by bending his abdominal tip toward hers. Male moths were found to produce ultrasonic courtship songs of extremely low intensity (46 dB peSPL at 1 cm) during these copulation attempts (Nakano et al., 2006; Nakano et al., 2008). Since the use of quiet sound would be advantageous in avoiding eavesdropping by predacious bats and rival males, this result prompted us to predict that whispering of ultrasonic courtship songs is widespread among moths (Nakano et al., 2008; Nakano et al., 2009). In the present study, we tested our prediction by examining 13 tympanate moth species, belonging to diverse taxonomic groups, for the production of quiet ultrasonic courtship songs.

We investigated the emission of courtship ultrasound in 13 tympanate moth species belonging to four families (three superfamilies): five species of Noctuidae (Noctuoidea), the cotton bollworm Helicoverpa armigera Hübner, the cabbage armyworm Mamestra brassicae Linnaeus, the common cutworm Spodoptera litura Fabricius, Herminia tarsicrinalis Knoch and Diomea cremata Butler; three species of Arctiidae (Noctuoidea), the tiger moth Spilosoma punctarium Stoll, the fall webworm Hyphantria cunea Drury and Eilema japonica Leech; one species of Geometridae (Geometroidea), the Japanese giant looper Ascotis selenaria cretacea Butler; and four species of Crambidae (Pyraloidea), the rice stem borer Chilo suppressalis Walker, the mulberry pyralid Glyphodes pyloalis Walker, Palpita nigropunctalis Bremer and Spoladea recurvalis Fabricius (see supplemental Table S1 for details).

Larvae of H. armigera, M. brassicae, S. litura, H. tarsicrinalis, E. japonica and A. selenaria were reared on a commercial artificial diet for silkworms, Silkmate™ 2M (Nosan Corp., Yokohama, Japan), and the other larvae were reared on their host plants under conditions of 24°C, 50-70% relative humidity, and a 16 h:8 h light:dark cycle. Pupae and/or adults were separated by sex, based on the genital morphology of the terminal abdominal segment, and maintained under the same environmental conditions as for rearing. Emerged female and male moths were transferred to 430 ml plastic cups with a supply of water until used.

A single unmated male and five to ten 1- to 4-day-old virgin females were introduced into a cubic mesh cage (18 cm×18 cm×18 cm) placed in a soundproof box (40 cm×40 cm×70 cm) with one side open, and maintained for >1 h before a recording. The activities of the insects were observed under a dim red light (0.2 lux). Sound recordings were made when the moths showed high mating activity, i.e. during the last 3 h of the scotophase. The exception was C. suppressalis, which showed high mating activity during the early scotophase (Samudra et al., 2002). Courtship sounds of male moths were recorded with a 1/4-inch (6.4 mm) condenser microphone [type 4939 (flat response from 3 kHz to 100 kHz with ±1 dB), Brüel & Kjær, Nærum, Denmark], the tip of which was placed 1 cm from the male. Signals from the microphone were amplified by pre- and conditioning-amplifiers [type 2670 and 2690 with a 0.02-100 kHz band-pass filter (−40 dB/decade), Brüel & Kjær]. The signals were digitized at a sampling rate of 300 kHz using a 14-bit A/D converter (Wavebook 512A, IOtech, Ohio, USA), and analyzed using the software BatSound 3.31 (Pettersson Elektronik AB, Uppsala, Sweden) after high pass filtering (>10 kHz, Butterworth filter). Power spectra of the sounds were computed using a Hanning window with an FFT (fast Fourier transformation) size of 1024 points. The peak equivalent sound pressure level (dB peSPL re. 20 μPa root mean square) of the sound was determined from the peak amplitude of the signal with reference to the signal of a sound level calibrator (type 4231, Brüel & Kjær; 94.00±0.20 dB SPL at 1 kHz) (Stapells et al., 1982).

To test our prediction that ultrasound signals used for courtship are less intense than those used in other contexts such as bat defense or mate attraction, sound pressure levels of moth ultrasounds reported to date were compared between two groups classified by the distance of communication, i.e. ‘close’ (within 5 cm) and ‘far’ (beyond 5 cm). Based on this criterion, species exhibiting close communication include Galleria mellonella (Spangler, 1985), Plodia interpunctella (Trematerra and Pavan, 1995) and Ostrinia furnacalis (Nakano et al., 2008), whereas species with far communication include Rileyana (Thecophora) fovea (Surlykke and Gogala, 1986), Amyna natalis (Heller and Achmann, 1993), Hecatesia exultans (Alcock and Bailey, 1995), Heliothis zea (Kay, 1969), Amphipyra perflua (Lapshin and Vorontsov, 2000), Syntomeida epilais (Sanderford and Conner, 1995), Cycnia tenera (Barber and Conner, 2006), Arctia caja (Surlykke and Miller, 1985), Euchaetes egle (Simmons and Conner, 1996), Phragmatobia fuliginosa (Surlykke and Miller, 1985), Pyrrharctia isabella (Krasnoff and Yager, 1988), Euchaetes bolteri (Simmons and Conner, 1996), Empyreuma affinis (Sanderford et al., 1998), Bena bicolorana (Skals and Surlykke, 1999), Pseudoips prasinana (Skals and Surlykke, 1999), Corcyra cephalonica (Spangler, 1987), Achroia grisella (Spangler, 1984), Symmoracma minoralis (Heller and Krahe, 1994), and 36 tiger moths described by Barber and Conner (Barber and Conner, 2006). Differences in the sound pressure levels emitted by the two groups were analyzed with a generalized linear model (GLM) using the software package R, version 2.7.0.

To verify that the sounds were used for sexual communication in S. litura, we examined how mating was affected by deafening females. For the operation, 0- or 1-day-old virgin females were anesthetized with CO₂, and the tympanic membranes on both sides of the metathorax were punctured using a fine insect pin under a stereomicroscope. The sham operation involved puncturing another part of the metathorax. Behavioral experiments were conducted in the last 2 h of the scotophase 1 day after the operation. A single 1-day-old intact unmated male was introduced into a cubic mesh cage (25 cm×25 cm×25 cm), which housed 5-10 deafened, sham-operated, or intact females. Multiple females were housed in the cage so that always, at least one female would be releasing sex pheromone during the experiment. In response to the sex pheromone released by the females, the male usually readily flew up to one of the females, and started courting. The observation of mating behavior was continued until the female accepted the male or rejected him by flying away. The emission of ultrasounds by the male was continuously monitored with an ultrasound detector (model D240x, Pettersson Elektronik AB). The pair observed was removed from the cage, and the experiment was continued with a new male. The males that did not show courtship behavior within 5 min from the introduction were removed, and these were not included in the data.

In nine of 13 moth species examined, we detected the emission of ultrasounds by males at a specific step in a series of courtship behaviors. Here, it should be mentioned that sounds specifically emitted in association with courtship could be distinguished from those produced incidentally with wing beats or the movement of other body parts by the distinctly high energy level in the ultrasound range (>20 kHz) of the courtship sounds.

Males of S. litura and H. tarsicrinalis produced courtship ultrasounds (Fig. 1), whereas males of three other noctuid species, H. armigera, M. brassicae and D. cremata, did not produce sounds in the ultrasound frequency range (>20 kHz).

Males of S. litura produced a series of bursts of ultrasonic clicks during courtship (the lower oscillogram in Fig. 1A). The sound was broadband high frequency, ≈20-80 kHz, with a peak power frequency of 51 kHz and a sound level of 70±2 dB peSPL at a distance of 1 cm (mean ± s.d., N=15). The bursts lasted 9.5-16.0 ms and numbered ≈60 per second whereas the clicks lasted 0.1 ms and numbered ≈1900 per second. The number of clicks in one burst was 36±14 with fewer in the beginning of song trains (upper oscillogram of Fig. 1A).

In H. tarsicrinalis, males emitted trains of pulses with a duration of ≈6.5 ms (Fig. 1B shows a single train of pulses in the upper oscillogram, three pulses in the middle oscillogram, and a single pulse in the lower oscillogram). Single pulses were repeated at a frequency of around 26 pulses per second. Each pulse consisted of 50-70 irregular clicks. Power spectra of the clicks showed energy in a broad frequency range up to above 100 kHz with a peak power frequency of 88.8 kHz (64±2 dB peSPL, N=2). The intensity of the clicks increased gradually from the beginning to end of the pulse.

Males of S. punctarium and E. japonica emitted ultrasounds during courtship (Fig. 2), whereas males of H. cunea did not.

Males of S. punctarium produced a regular pattern of pairs of ultrasonic clicks (Fig. 2A). A single click lasted 0.4-0.6 ms, and was repeated at 10-100 clicks per second. There was large individual variation in the sound level (66±4 dB peSPL, N=12), but the first click of the pair (63±5 dB peSPL, N=12) was always significantly weaker than the second (Wilcoxon signed rank test, N=12, V=78, P=0.0025). The peak power frequency of the clicks was around 38 kHz.

When close to a female, courting males of E. japonica emitted bursts of ultrasonic clicks with two to six, predominantly four, clicks in a single burst (Fig. 2B). One burst lasted 50-85 ms, and was repeated ca. 1.2 times per second. The clicks had a peak power frequency of 41 kHz. Clicks lasted ≈2 ms with a repetition rate of 35-70 clicks per second (N=13). The first, third and fifth clicks of one burst were significantly more intense (62±3 dB peSPL at 1 cm) than the second, fourth and sixth clicks (56±3 dB peSPL; Wilcoxon signed rank test, N=13, V=91, P=0.00024).

Males of A. selenaria produced bursts of clicks with a peak power frequency of 43 kHz and a maximum sound intensity of 73±3 dB peSPL (N=9; Fig. 3). One song train lasted 200-600 ms and contained 10-30 bursts. One burst consisted of 4-6 clicks and lasted 6-10 ms.

Males of all crambid moths tested, i.e. C. suppressalis, G. pyloalis, P. nigropunctalis and S. recurvalis emitted courtship ultrasounds (Fig. 4).

C. suppressalis males emitted ultrasounds with a peak power frequency of 52 kHz (76±3 dB peSPL, N=8; Fig. 4A). A single train (upper oscillogram of Fig. 4A) was composed of pulses (middle oscillogram of Fig. 4A) lasting ≈4 ms and repeated at 40-55 pulses per second. The number of clicks in single pulses ranged from 7 to 15.

Males of G. pyloalis produced an irregular series of ‘hyper’ ultrasonic clicks with peak energy above 100 kHz. Clicks were repeated on average at 1075 clicks per second. The duration was ≈0.13 ms and the peak power frequency was 113 kHz. The maximum sound level was ≈54±2 dB peSPL at 1 cm (N=6; Fig. 4B).

P. nigropunctalis males emitted several bursts of clicks in one train (upper oscillogram of Fig. 4C). Single bursts lasting ca. 19 ms were repeated 16 times per second, and contained 10-14 clicks (middle oscillogram of Fig. 4C). The duration of single clicks within a burst was around 0.13 ms, the maximum intensity was ≈71±1 dB peSPL, and the bandwidth was broad, ranging from 30 to 120 kHz with a peak around 100 kHz (N=2; Fig. 4C).

Males of S. recurvalis emitted discontinuous clicks lasting 0.12 ms at a high rate, 1140 clicks per second. The maximum sound level was ≈43±3 dB peSPL, and the peak power frequency was around 115 kHz (N=7; Fig. 4D), hence with a bias toward hyper ultrasound over 100 kHz. Because frequencies over 100 kHz are beyond the specifications of the present recording system, maximum sound levels of G. pyloalis, P. nigropunctalis and S. recurvalis are likely to be underestimated.

Reported sound levels of the acoustic signals produced by various moths differed widely in maximum energy from 43 to 125 dB peSPL at 1 cm (Fig. 5A). The levels were not dependent on the mechanism of sound production or on the taxonomic group (family) to which the moth belongs [likelihood ratio (LR) test in generalized linear model (GLM), Gaussian error, identity link, mechanism: χ²_2,21=41.5, P=0.87; family: χ²_5,21=981.9, P=0.30]. The sound pressure levels of courtship ultrasounds were significantly lower than those of signals emitted for bat defense, mate attraction and territorial display (LR test in GLM, χ²_1,64=7396.9, P<0.0001; Fig. 5B).

In S. litura (Noctuidae), males emitted ultrasound in association with courting. Deafening of females demonstrated the necessity for male ultrasound since males had substantially higher mating success with hearing than with deaf females; all of the intact females (100%, N=15) and almost all the sham-operated females (93%, N=15) accepted the courting male, whereas only 52% of the deafened females (N=21) accepted the male.

To date, only a small number of moth species have been reported to produce ultrasounds for sexual communication (Conner, 1999; Greenfield, 2002). In the present study, we explored the possibility of weak ultrasound production by courting male moths. We showed that the production of weak ultrasounds is much more widespread among moths than previously assumed, because nine of 13 tested species produced sounds in close proximity to a female, after showing typical female sex pheromone-mediated mating behaviors (Yushima and Tamaki, 1974; Seol et al., 1986; Witjaksono et al., 1999; Samudra et al., 2002; Shirai, 2006).

The ultrasounds emitted by courting males varied extensively in temporal and spectral properties across species. The ultrasonic signals produced by males of S. litura, S. punctarium, E. japonica, A. selenaria, G. pyloalis, P. nigropunctalis and S. recurvalis consisted of single, often irregular clicks (Fig. 1A, Fig. 2, Fig. 3 and Fig. 4B-D), suggesting that these species have smooth tymbal organs as found in tiger and wax moths (Table 1) (Spangler, 1988; Fullard, 1992; Conner, 1999). By contrast, the sounds produced by H. tarsicrinalis and C. suppressalis were not single clicks but regularly generated pulses (Fig. 1B and Fig. 4A) characterized by consecutive clicks, suggesting that the two species have stridulating organs such as the file-and-scraper found in Rileyana fovea (Noctuidae) (Surlykke and Gogala, 1986) and Syntonarcha iriastis (Crambidae) (Gwynne and Edwards, 1986) (Table 1). With respect to spectral properties, the frequency of maximum energy produced by males ranged broadly from 38 to 115 kHz (Table 1). All moths studied to date produce ultrasounds of less than 100 kHz except for Achroia grisella (Pyralidae) (Conner, 1999). It is worth noting that some crambid moths produced ‘hyper’ ultrasonic frequencies (over 100 kHz), a frequency range that only a few bats use for echolocation (Miller and Surlykke, 2001; Schnitzler and Kalko, 2001). Thus, the results indicate that moths belonging to different genera and families have evolved various mechanisms to emit ultrasounds during courtship (Table 1).

Our findings corroborate the prediction that many moths are not silent but emit quiet ultrasounds, probably keeping the sound level low to avoid eavesdropping by predators, parasites and rival males (Nakano et al., 2008; Nakano et al., 2009). Here we provided evidence for acoustic sexual communication in S. litura, since our results clearly showed that mate acceptance by females was much higher when she could hear the ultrasound produced by the male. We are continuing behavioral studies and have preliminary evidence that in some moth species examined in the present study, ultrasound does function in acoustic sexual communication.

We have suggested that a large proportion of moths communicate acoustically at close range using quiet ultrasound, which is audible to their mates but inaudible to unintended receivers. Moreover, the use of hyper-frequency ultrasonic signals is perhaps a good strategy to avoid detection by enemies because such high-frequency sounds are not utilized by most common predators of moths. This strongly suggests that acoustic communication using quiet ultrasound is much more widespread among hearing moths than hitherto presumed because it is so inconspicuous. We found that 70% of the moths we examined produced ultrasound, which suggests that in moths, close-range sound communication may be the norm rather than the exception, which has important consequences for our understanding of the co-evolution of bat and moth audition.

We thank S. Tatsuki, S. Hoshizaki and two anonymous reviewers for valuable comments, and H. Kitajima, R. Sakai, T. Honda, K. Murata and A. Yamasaki for collecting insects.