Behavioral Response to Ultrasound by The Tiger Beetle Cicindela Marutha Dow Combines Aerodynamic Changes and Sound Production

The broad range of insects with ultrasonic hearing in the range 30–60 kHz attests to the ecological pressure exerted on them by echolocating, insectivorous bats (reviewed by Hoy et al. 1989). While some of these insects also use their hearing for intraspecific communication, its use in bat evasion has been substantiated in many species by establishing that bat-like trains of ultrasonic pulses trigger specific behaviors in flying animals that are unique to that behavioral context. In two cases (moths and mantises), behavioral observations in the field have demonstrated the efficacy of the insects’ ultrasound-triggered responses in evading capture by hunting bats (Roeder, 1967; Yager et al. 1990).

The ultrasound-triggered behaviors vary both in complexity and in type, although most involve wingbeat pattern or rate changes. Green lacewings Chrysopa carnea (Miller and Olesen, 1979) and tettigoniids Neoconocephalus ensiger (Libersat and Hoy, 1991) respond by closing their wings and dropping, locusts Locusta migratoria (Robert, 1989) and crickets Teleogryllus oceanicus (Moiseff et al. 1978) employ specific steering motions, and mantises Parasphendale agrionina extend their forelegs and dorsiflex the abdomen (Yager and May, 1990). Among the moths, the noctuids vary their response according to stimulus intensity (=bat range), turning away from weak signals and diving erratically from more imminent danger (Roeder, 1967). The arctiids answer the threat with a unique strategy: flight path changes are much less common than in other moths, but they consistently produce trains of ultrasonic clicks that are very effective in averting the bat’s attack (Dunning and Roeder, 1965; Dunning et al. 1992; Acharya and Fenton, 1992). Startle, disruption of the bat’s echolocation system and advertisement of the moth’s distastefulness are mechanisms that singly, or in combination, may explain how the clicks protect the moths (Fullard et al. 1994).

Behavioral and physiological experiments have recently established that some tiger beetles (Cicindelidae) hear using two ears located on the dorsal surface of the first abdominal segment (Spangler, 1988a; Yager and Spangler, 1995). Afferent nerve recordings show that the auditory system has a sensitivity of 50–60 dB SPL in its most sensitive frequency range of 25–35 kHz. The function of hearing in these beetles is not yet known. Since they produce sounds with a peak frequency of 30–35 kHz in non-defensive contexts, intraspecific communication is a possibility (Freitag and Lee, 1972; Pearson, 1988). However, an alternative hypothesis is also viable: tiger beetles fly at night, as shown by frequent black-light captures of many species (Larochelle, 1977; Pearson, 1988) and coexist with many species of insectivorous bats (Hoffmeister, 1986), so their auditory system may help them avoid capture by bats. These two hypotheses are not mutually exclusive.

The present study lends support to the hypothesis of ultrasound-mediated bat evasion by tiger beetles by demonstrating a complex behavioral response to pulsed ultrasound that occurs only during flight. Furthermore, the behavioral response is unique in that it combines both wing motion changes (presumably altering the flight path) and the production of loud ultrasonic clicks.

We collected adult male and female Cicindela marutha Dow (Coleoptera: Cicindelidae) near Willcox Playa, approximately 137 km east of Tucson, AZ, USA, during late July and August. We conducted experiments both in Tucson and in College Park. Some beetles were held in the laboratory at 24–28 °C for up to 4 weeks in sand-filled aquaria containing puddles of standing water; day length was 12 h. They were fed small crickets, flies and insect parts several times a week.

Experiments were carried out either in an acoustic isolation chamber (Industrial Acoustics Corp.) or at the volume center of a 1 m×1 m×2 m box lined with acoustic foam to minimize echoes. We suspended each beetle from a thin wire affixed to its pronotum with a small drop of wax. For experiments in the anechoic box, the position of the beetle was standardized at 0.7 m from speakers (Technics EAS10TH400B) at either end of the box. In the acoustic chamber, stimuli were broadcast from Realistic 40-1375 leaf tweeters 0.4–0.5 m away from the beetle or from a custom-built bat simulator (40 kHz, 10 ms pulses at 10 s⁻¹) 0.15 m away from the beetle, depending on the experiment.

Stimuli were trains of 10 ms trapezoidal sound pulses (0.5–3.0 ms rise/fall times for different experiments; 10 or 20 ms interpulse interval) produced by standard electronics. The rise/fall times were adequate to eliminate onset and offset transients from the speaker in all cases. Sound pressure level (SPL) was calibrated from 1 to 100 kHz using a Brüel & Kjaer 2230 SPL meter and 4135, 6.25 mm microphone (grid off). Sound pulses used for calibration were longer than 100 ms to allow adequate meter response time. The calibration system frequency response was flat to within 1 dB from 1 to 70 kHz; appropriate corrections were included in the calibration procedure to compensate for a roll-off above 70 kHz. Stimulus SPLs are reported here as dB (RMS) re 20 μPa (dB SPL). Our calibration confirmed attenuator accuracy over the range used and speaker linearity up to 95 dB for frequencies ⩽ 80 kHz. Harmonics with the greatest energy (usually the second) were more than 35 dB (most were more than 45 dB) below the fundamental at all frequencies (Stanford Research Systems SR760 FFT analyzer).

Tethered beetles initiated flight either spontaneously or in response to puffs of wind. Sustained flight did not require continuous wind, and we did not use a fan during data collection. The data are from more than 50 beetles. Temperatures during testing were 22–26 °C.

We monitored the sound produced during normal flight and after an ultrasonic stimulus using a Brüel & Kjaer 4135 6.25 mm microphone with the Brüel & Kjaer 2230 SPL meter as an amplifier. Unless otherwise noted, the microphone pointed at the wings from 2 cm behind the animal and 30–45 ° above the horizontal. A consistent pressure wave is produced during each cycle of wing motion, so we also used this arrangement to measure wingbeat frequency (we confirmed these measurements using the shadow cast on a photocell by the hindwing as it broke a laser beam). For spectral analysis, the signal was stored on a Racal IV-D tape recorder running at 38 cm s⁻¹ (recording system frequency response was flat to within 2 dB up to 100 kHz). For experiments not requiring acoustic analysis (presence or absence of clicks, measurement of wingbeat frequency, etc.), we stored the data on either a Marantz PM-455 cassette recorder (±2 dB to 16 kHz) or a BioLogic digital instrumentation tape recorder (±1 dB to 20 kHz).

For click SPL measurements, we recorded 30 kHz reference tones of known SPLs between 65 and 95 dB SPL on the tape (Racal recorder) before recording ultrasound-evoked clicks on the same tape at identical gain. Click SPLs were then derived by comparison of click versus reference tone output voltages on an oscilloscope.

We used photographic techniques to assess changes in wing position and movement pattern after ultrasonic stimulation (tripod-mounted Nikon N90 camera with Nikkor 105 mm macro lens; Kodak TMAX 100 film). By choosing an appropriate shutter speed, we could either freeze motion (shutter speed faster than 0.01 s) or obtain an ‘average’ position over approximately 25 wingbeat cycles (shutter speed 0.5 s). Four different photographic series optimized viewing angle for measurement of specific aspects of elytra or hindwing change. Stimuli were 500 ms trains of 10 pulses at 30 kHz and 15 dB over threshold. Data were collected and analyzed in a paired design using photographs immediately before and after each stimulus.

We established the latency from stimulus onset to first elytral motion by first aiming a 0.5 mm laser beam from above so that it struck the extreme leading edge of the opaque elytrum about half-way along its length when the beetle was in stable flight. As the elytrum moved backwards after stimulation, it exposed a photocell positioned below the animal to the laser beam. Voltage output of the photocell and a stimulus monitor were stored on separate channels of the digital tape recorder.

Stimuli used to determine the tuning for components of the ultrasound-triggered response were 300 ms pulse trains (10 ms pulses; 10 ms interpulse interval) with intertrain intervals of greater than 5 s to minimize habituation. Frequencies were presented in a semi-random order so that adjacent frequencies did not occur in series. Threshold was defined as the lowest SPL eliciting the response in at least two out of three trials.

We used a Kay Elemetrics DSP Sona-Graph 5500 in its 32 kHz range for analysis of the beetle-produced clicks. The signals on the data tapes were reproduced at one-quarter of the recording speed on the Racal tape recorder, allowing analysis to cover the entire frequency range of the recording.

Analyses of temporal patterns and latency utilized data acquisition and analysis software, SuperScope II (GW Instruments, Somerville, MA, USA), running on a Power Macintosh computer (Apple). Analog-to-digital conversion (MacAdios II board; GW Instruments) used sampling periods of 10 μs or above, depending on the analysis.

We obtained angular measurements of the elytra and hindwing movements from the four series of photographs by first scanning the contact sheets of each roll of film into the Power Macintosh computer (La Cie Silverscan scanner controlled by ColorIt! 3.0 software). Image enhancement and analysis tools were provided by NIH Image software. In some cases, measurements were made on images projected from negatives by a photographic enlarger. We estimate that the overall resolution of the analysis was ±1 °. We define 0 ° in the horizontal plane as directly in front and 0 ° in the vertical plane as directly above the animal. We used circular statistics for computations dealing with angular measurements and a non-parametric test for angular comparisons (sign test) to avoid assumptions about the underlying data distribution (Batschelet, 1981).

For determining mean tuning curves, data in decibels were converted to a linear scale (pressure) prior to statistical manipulation and the results were then converted back to decibels for reporting. This explains the unequal lengths of the standard deviation bars above and below data points on the figures.

Averages are expressed in the text as mean ± standard error (S.E.M.) for linear data and mean ± angular standard deviation (S.D.) for angle data unless otherwise noted. The significance level for all statistical tests is 0.05.

C. marutha flies strongly on a tether without any wind stream, and many individuals only showed obvious signs of fatigue after 30–45 min of continuous flight. Under our conditions, females flew more reliably than males, but we detected no other behavioral differences between the sexes in our tests. We noted a consistent time-of-day effect: the beetles flew most readily in the morning and evening and were reluctant to fly during mid-afternoon.

Tiger beetles do not flap their elytra. The only movement of these thickened forewings is a slight vibration arising from the action of the hindwings transmitted through the thorax. In normal, stable flight, the elytra are held symmetrically to the sides at 97±7 ° (with 0 ° to the front of the animal; 254 measurements on 10 beetles) relative to the long axis of the body (e.g. Fig. 1A). They are raised above the horizontal by 21±6 ° (226 measurements on 10 beetles), forming a dihedral (a shallow ‘V’ shape; Fig. 2A).

The hindwings flap at a mean of 45.9±0.4 strokes s⁻¹ (Hz; 156 measurements from five beetles) in stable, tethered flight at 22 °C. Eleven beetles flying under the same conditions, but at a higher temperature (26 °C) had an average wingbeat frequency (WBF) of 52.8±0.9 Hz. Among all beetles, the range of WBFs was from 34 Hz to a maximum of 57 Hz. The WBF for an individual beetle during a testing period was not constant, but varied by 2–3 Hz around the mean.

The hindwing excursion in stable, tethered flight is 167±9 ° (see Fig. 2; 226 measurements on 10 beetles). The two wings almost touch at the top of the stroke (90 ° to the horizontal); the bottom of the stroke is 77 ° below the horizontal. The hindwings do not beat vertically (see Fig. 3). The average stroke plane angle shown by the 0.5 s photographic exposures of beetles viewed from the side is 118±8 ° (53 measurements on five beetles) relative to a line through the head and thorax; the top of the stroke is further caudal than the bottom. These photographs do not allow distinction between the angles of the downstroke and the upstroke.

During stable flight, the tiger beetle’s head is not tilted to either side (Fig. 2). The abdomen droops somewhat below the line defined by the head and thorax (Fig. 3). The prothoracic legs are partly extended to the front, and the metathoracic legs trail straight behind the animal (Figs 1, 3). The mesothoracic legs are also fully extended and kept tight against the lower surface of the elytra (possibly held in a groove); the tarsi extend past the elytral tips (Fig. 1).

Flying C. marutha react to ultrasound with a multiple-component behavior (Figs 1–3). Within 150 ms, the beetle (1) rolls its head to the side, (2) deflects one or both metathoracic legs to the side, (3) swings the elytra backwards and pronates them, and (4) increases hindwing stroke frequency and excursion, and changes the stroke plane angle. It often also contracts the dorsal abdominal muscles, causing the abdomen to compress forward slightly. In some cases, the prothoracic legs also swing to the side. The response is context-specific: it occurs only during flight. In fact, we did not observe any reaction to ultrasound by beetles walking or standing on the substratum in more than 10 trials.

During stable flight, the beetles hold the elytra approximately at right angles to the body (97 °). The rapid posterior swing of the elytra triggered by ultrasound (Figs 1, 3) moves them to an average angle of 103±7 °, a statistically significant change (sign test; 10 animals). The maximum swing we observed was 23 °, and swings of 10–20 ° were frequent. It is not uncommon, especially soon after flight onset, for the beetles to swing the elytra all the way back, closing them and halting flight. The beetles also often lower the elytra slightly so that the dihedral viewed from the front becomes shallower (Fig. 2). The mean dihedral angular change, however, is small (from 21±6 ° to 17±7 °) and is not statistically significant. The shallower dihedral may actually reflect pronation in some cases rather than, or in addition to, a change in elytra position. In many trials, ultrasound triggered a clearly visible tilting of the elytra so that the leading edge dropped towards the horizontal. While the pronation appeared to be symmetrical, our photographs do not allow us to make the angular measurements necessary to confirm this.

As shown in Fig. 4, latencies to the first movement of the elytra are less than 110 ms for SPLs of 10 dB over threshold or higher. The shortest mean latency we measured in any individual beetle was 78.7 ms at 25 dB over threshold. The latency rises very steeply at lower SPLs; the one measurement obtained at 2 dB over threshold (data not shown) was longer than 300 ms.

The elytra swing component of the behavior is broadly tuned (Fig. 5). Greatest sensitivity is at 30 kHz, but the response is only slightly less sensitive at 60 kHz. The lowest individual threshold value we recorded was 68 dB SPL at 30–35 kHz. The sensitivity drops off very sharply below 20 kHz.

The hindwing component of the ultrasound-triggered behavior is itself complex. Wingbeat frequency increases in all cases, and the duration of the increase usually exceeds that of the stimulus (e.g. Fig. 6C), although the response duration can be shorter that the stimulus duration for SPLs just above threshold. The magnitude of the WBF increase depends on the WBF just before the stimulus so that the beetle always responds to an ultrasonic stimulus with a maximum WBF in the range 50–60 Hz (Fig. 7). For most of our beetles, the increase was 6–10 % (3–5 Hz). Considering only pre-test WBFs of 43 Hz or greater, the slope of the linear regression is significantly different from zero (F-test; d.f.=47) for both the percentage increase (1.44 % decrease per Hz increase in pre-test WBF; r²=0.294) and the maximum WBF (0.46 Hz increase in maximum WBF per Hz increase in start WBF; r²=0.172). We found no relationship between the SPL of the stimulus and the magnitude of the WBF change. In some cases, the increase in WBF was preceded by a decrease lasting for 2–3 cycles.

Hindwing stroke excursion also increases in response to ultrasound (Fig. 2). The control angle of 167±9 ° measured from the front-view photographs changes to 172±8 ° after stimulation, a statistically significant difference (sign test; 10 animals). As for WBF, the magnitude of the change depends on the pre-test condition: beetles with smaller initial excursion angles showed greater increases so that the response angle was always close to 180 °. Even though the hindwings normally reach 90 ° to the horizontal at the top of the stroke, we saw the increased excursion reflected there as well. Observations using a stroboscopic light suggested that the wingtips may actually touch at the top of the stroke following a stimulus. The hindwing excursion increase substantially outlasts the stimulus (Fig. 6A).

The stroke plane angle of the hindwing relative to the body after stimulation is difficult to measure with reasonable precision in our 0.5 s exposure, side-view photographs because the anterior wing borders become blurred compared with the pre-test exposure. Observations from the photographs suggest that the tops of the wings are further forward after stimulation (see Fig. 3). However, the actual angular difference measured is less than 2 °, which is not a statistically significant change (sign test; five animals). We argue, nevertheless, that a change in the hindwing path is a real component of the overall behavior on the basis of two observations: (1) our front-view photographs reinforce the suggestion of a change in the stroke plane of the hindwings (the overall reflection of light from the hindwings differs before and after the stimulus) (Fig. 2); (2) in strongly responding animals, the entire posterior half of the animal appears to tilt forwards in response to an ultrasound stimulus, possibly the combined visual effect created by a change in the stroke plane angle of the hindwings and contraction of the dorsal abdominal musculature (Fig. 3).

The pattern of latency change with increasing SPL is very similar for the changes in WBF and hindwing excursion after stimulation (Fig. 4). Minimum mean latencies for individual beetles were 75 and 79 ms for excursion and WBF, respectively. In contrast with the elytra swing, however, hindwing changes show relatively smaller latency increases as SPL decreases towards threshold.

We find no convincing evidence that the ultrasound-triggered response is consistently oriented relative to the sound source. We tested 10 beetles by presenting 30 kHz stimuli at 90 dB SPL alternately from 90 ° to the left and right. Each of the 20 trials for each beetle was followed by a disruptive stimulus (wind puff or light flash) from straight ahead that returned the animal to a symmetrical flight posture. Only three of the animals showed a statistically significant bias (binomial probability) in the direction of the head roll; the top of the head turned towards the stimulus more often than away from it. Pooling the data from the 10 beetles yields a small, but significant (χ²), difference: towards the stimulus in 115 trials and away in 86. Five of the animals showed a significant bias (binomial probability) in the direction of the leg kick, but two kicked mostly towards the stimulus and three away from the stimulus. The pooled data indicate no preferred direction (χ²) for the leg kick.

While not consistently directional relative to the sound source, the response is clearly lateralized. For instance, the legs almost always (171 out of 180 trials) kick in the direction opposite to the head roll. It is especially easy to appreciate the lateralization when observing reactions to longer (0.3–2 s) ultrasonic stimuli at 95–105 dB: after an initial strong response, many beetles ‘waggle’ – the head, legs and wings appear to swing from side to side in a coordinated manner.

The behavioral response of flying C. marutha to ultrasonic pulses also includes a prominent acoustic component. The beetles produce trains of loud, ultrasonic clicks (Fig. 8A).

The individual clicks produced by the beetles are very short (Fig. 8C). The major portion of the click is over within 150 μs; a low-amplitude tail lasts for an additional 100–150 μs. The spectral energy is distributed over 10–70 kHz, with a broad peak at 30–40 kHz (Fig. 8B). There is sufficient energy below 15 kHz for the clicks to be faintly audible to the human ear. Normal wingbeat sounds have virtually all of their energy below 8 kHz and do not include click-like components.

The average maximum SPL of the clicks from five beetles (9–16 trials each) was 89.6 dB SPL (peak) (+4.3 dB, −8.4 dB, S.D.) with a range of 81–95 dB SPL; the microphone was pointed at the animal from 2 cm behind and approximately 30 ° above the horizontal. The wingbeat sound measured just before each stimulus averaged 69.4 dB SPL (peak) (+2.3 dB, −3.2 dB, S.D.). In four beetles, we compared the maximum SPL of the clicks recorded from behind and from a symmetrical position ahead of the animal. In all cases, the clicks were louder from the front (t-tests; d.f.=19–49). The average difference was 5.2 dB.

Using 1 s trains of 10 ms pulses (20 ms period) at 30 kHz and 90 dB SPL, we assessed click production in seven beetles. Each stimulus train elicited an average of 78.9±8.6 clicks (N=52). Weakly responding beetles produced fewer than 20 clicks per stimulus, whereas strong responders routinely produced 150–190. Clicking greatly outlasted the stimulus in all cases (Figs 6B, 8A).

Both the latency to the first click and the number of clicks produced varies with SPL. As for the elytra swing, the latency decreases very rapidly as SPL increases above threshold (Fig. 4). The shortest mean latency we measured for an individual beetle was 98.1 ms at 20 dB over threshold. The number of clicks per 500 ms stimulus train follows a similar (but inverse) pattern (Fig. 9). The clear plateau suggests a dynamic range of approximately 20 dB.

The behavioral tuning curve for click production matches the curve for elytra swing very closely (Fig. 5). The lowest individual threshold value we observed was 65 dB SPL at 30 kHz.

Click production is clearly linked to the wingbeat cycle. As Fig. 10 shows, the clicks occur at the wingbeat frequency. The primary, high-amplitude click occurs just before the wingbeat waveform in every cycle. (The wingbeat waveform is a pressure wave lasting 5 ms or less that was recorded by the microphone once each wingbeat cycle whenever the wings are flapping. We do not know its exact relationship to wing position.) When a smaller secondary click is present (as in three cycles in Fig. 10), it also has a constant temporal relationship to the wingbeat waveform. The second click typically occurred 4–6 ms after the first. For 2301 cycles, seven beetles produced an average of 1.78 clicks per cycle. While the majority of cycles had one or two clicks, 21.2 % of the cycles had three or more clicks (maximum of six; generally low in amplitude compared with the primary click). High click densities were limited to strongly responsive animals and to the portion of the click train during the stimulus in moderate responders. The ‘extra’ clicks were not scattered randomly throughout the wingbeat cycle, but clustered with the primary and secondary clicks.

Tiger beetles did not produce clicks in response to tactile stimuli, even when flying.

Trains of ultrasonic pulses elicit a complex behavioral response in flying C. marutha. Changes in wingbeat pattern and movements of the head, legs and elytra are coupled with production of loud ultrasonic clicks. This combination of behavioral components is unique among insects responding to ultrasound. While both the phylogeny and the distinctive structure of the tiger beetle auditory system ensure that this behavioral suite evolved independently, some of the components show striking similarities to acoustic behaviors in other insects.

Ultrasound triggers defensive behaviors in green lacewings Chrysopa carnea (Miller, 1975; Miller and Olesen, 1979), tettigoniids Neoconocephalus ensiger (Libersat and Hoy, 1991), moths (reviewed by Roeder, 1967; Spangler, 1988b), crickets Teleogryllus oceanicus (Moiseff et al. 1978; Nolen and Hoy, 1986), locusts Locusta migratoria (Robert, 1989) and mantises Parasphendale agrionina (Yager and May, 1990) as well as in tiger beetles. In all cases, context is crucial in determining the nature of the response and, in fact, most insects do not respond at all unless they are flying.

With the exception of the elytra swing, the non-acoustic components of the tiger beetle’s response are also found in various combinations in the other ultrasound-responsive insects. For example, crickets, locusts and tiger beetles share a directional leg kick. In crickets and locusts, there is a directional abdominal bending, but in mantises and tiger beetles there is a symmetrical contraction of the dorsal muscles. WBF changes are common to all, but mantises decrease the WBF while lacewings and tettigoniids stop flapping altogether. The head roll is absent in tettigoniids and possibly in lacewings.

These insects also share important functional characteristics of their ultrasound-triggered responses, presumably reflecting convergence driven by the shared problem of bat predation. In every case, the behavioral tuning curves are broad, spanning the range 30–60 kHz. Minimum thresholds range from 40–50 dB SPL (locusts, moths) to 70–80 dB SPL (tiger beetles). Latencies are uniformly short: less than 150 ms even for complex behaviors and as low as 25–35 ms for electromyographic events in muscles controlling specific components of the response (crickets, tettigoniids). The duration of the response substantially exceeds that of the stimulus in each case.

Arctiid moths and tiger beetles are the only insects known to click in response to ultrasound in a defensive context. In both cases, the clicks are very short (50–200 μs) and have a broad (30–80 kHz) frequency spectrum (Fullard and Fenton, 1977). Click SPL is also closely comparable: Krasnoff and Yager (1988) and Surlykke and Miller (1985) measured peak SPLs at 2 cm of 79–90 dB SPL for three arctiid species. Whereas C. marutha clicks at an average rate of 100–120 clicks s⁻¹, click rates vary among arctiid species from less than 20 to greater than 1500 clicks s⁻¹ depending largely on the structure of the tymbals that produce the clicks (Fullard and Fenton, 1977; Surlykke and Miller, 1985). The arctiid response to ultrasound includes some behavioral components other than clicking (WBF changes and flight cessation; Dunning and Roeder, 1965; Fullard, 1979), but in recent field studies (Dunning et al. 1992; Acharya and Fenton, 1992) nine species of arctiids known to click copiously survived bat attacks without performing any obvious evasive maneuvers in more than 80 % of observations (in control experiments, deafened and/or muted moths were captured in more than 60 % of the trials). In contrast to tiger beetles, arctiids will click in response to a range of threatening stimuli, whether in flight or not.

Some of the insects discussed here do not predictably orient their response relative to the ultrasound source (mantises, lacewings, tettigoniids), while others show strong directionality (moths, crickets, locusts). Tiger beetles appear to fall in the former category. Yager and Spangler (1995) showed that directional information is available to the central nervous system in the afferent responses, but also reported considerable variability.

Recently, Forrest et al. (1995) reported an ultrasound-triggered response in a scarab beetle. The comparison with cicindelids is especially interesting since these two coleopteran auditory systems must have evolved independently (T. G. Forrest, personal communication). Their study focused on the head roll shown by walking scarabs in response to ultrasonic pulses. The tuning was broad (20–70 kHz) at levels of 60–70 dB SPL. Latency to response in the neck muscles measured electromyographically was 30–40 ms. Of particular note is the lack of context-specificity: the scarabs responded both when walking and when flying. The in-flight response, however, has not yet been characterized.

The results presented in the present study may help to resolve the question raised by Yager and Spangler (1995) regarding how tiger beetles use their hearing. Two theories dominate: intraspecific communication and predator avoidance. Prey localization is unlikely since most of these insects are diurnal, visual hunters (Pearson, 1988). Yager and Spangler (1995) argued that both theories are possible, but that bat evasion was best supported by their physiological data.

The behavioral data presented here strongly support the hypothesis that hearing in C. marutha functions as part of a system for evading capture by hunting, echolocating bats. The context-specificity of the response clearly dictates a function during flight, and light-trap captures demonstrate that some tiger beetles (including C. marutha) fly at night in substantial numbers (nocturnal activities are primarily dispersal and, for some species, moving to oviposition sites; Larochelle, 1977; Pearson, 1988; D. L. Pearson, personal communication). Aerodynamically, the increases in hindwing excursion and WBF predict higher power output, and the elytra pronation, tilting forward of the stroke plane of the hindwings and forward compression of the abdomen would all contribute to a dive. The asymmetries in the head and leg movements suggest that the dive will be to the side, but non-static aerodynamic testing will be required to confirm this. The tuning of the behavioral response matches the power spectrum of the most common bat echolocation calls, as well as the audiograms of other insects known to use hearing to evade bats (reviewed by Yager and Hoy, 1989; Yager et al. 1990). Calculations using the behavioral threshold data for tiger beetles to determine the response time available to the insect when approached by a hunting bat (assuming beetle flight speed of 3 m s⁻¹, bat flight speed of 6 m s⁻¹ and echolocation cry SPL of 100 dB SPL at 10 cm; Yager et al. 1990) yield response times of 530 ms for a bat approaching from behind and 170 ms for a bat approaching from the front. Even in the worst circumstance, the behavioral latencies of less than 150 ms give the beetle adequate time to initiate evasive maneuvers before it can be captured.

The arctiid-like clicking of tiger beetles potentially adds another dimension to their defense against bats. It is well-established that arctiid moth clicks very effectively deter and/or disrupt bat attacks (Dunning and Roeder, 1965; Dunning et al. 1992; Acharya and Fenton, 1992). The exact mechanism is not clear and may vary depending on the circumstances. On the basis of the similarity of their power spectrum and SPL, tiger beetle clicks, like those of arctiids, will be readily audible to an insectivorous bat and could startle it (Bates and Fenton, 1990) or possibly interfere with its echolocation (Fullard et al. 1994). Also like arctiid moths, tiger beetles produce noxious defensive secretions (Pearson et al. 1988), so the clicks might deter attack by warning an experienced bat (aposematism). An intriguing possibility is that tiger beetles may be acoustic mimics (Batesian or Müllerian) of the distasteful, clicking moths.

The behavioral data also support the hypothesis that hearing in C. marutha is used for intraspecific communication. While the behavioral tuning curves for tiger beetle defensive behavior match closely those of other bat-evading insects, they do not match well the audiogram obtained physiologically from the tympanal nerve (Fig. 5; Yager and Spangler, 1995). The auditory afferent response is much more sharply tuned (to 30 kHz) and is more than 20 dB more sensitive. This implies a second function for hearing in tiger beetles. Acoustic signals produced by tiger beetles when on the ground are species-specific click trains with very different temporal patterns from defensive clicking (Freitag and Lee, 1972; H. G. Spangler, unpublished observations), reinforcing the idea that a second function may be intraspecific communication. This would be analogous to the situation in some crickets, for example, with broad ultrasonic tuning for evasion and low-threshold, sharp tuning, at a lower frequency, for mate attraction (Nolen and Hoy, 1986). The nature and function of intraspecific communication in C. marutha remain unknown.

Flying insects can produce ultrasonic clicks using a diversity of mechanisms. Arctiid moths use tymbals located on the sides of the thorax (Fullard and Heller, 1990). Agaristine moths have specialized ‘castanets’ on the forewings that strike each other at the top of the upstroke (Bailey, 1978), and some noctuid moths produce weak clicks using the same method, but without wing specializations (Waters and Jones, 1994). A nymphalid butterfly makes loud clicks when the tegula near the wing base hits a ‘costal clicker’ (a modified wing cell) on the forewing (Møhl and Miller, 1976). Motion of a projection on the forewing of pyralid moths deforms a modified tegula to produce clicks (Spangler and Takessian, 1986). A hard exoskeleton such as that of the tiger beetle provides innumerable opportunities for moving structures to strike or scrape together to yield clicks.

While we do not as yet know how tiger beetles produce their ultrasonic clicks, our data do provide come clues. Most suggestive is the tight temporal locking of the clicks to the wingbeat cycle. A mechanism not linked to the wings, such as the tymbals of arctiids, is therefore unlikely. The elytra do not flap, so any mechanism requiring rhythmic forewing movement can also be excluded. The power spectrum of the clicks shows a prominent peak at 30–40 kHz. This matches the power spectrum of ‘quacks’, acoustic signals of unknown function produced by C. marutha and many other tiger beetles (Freitag and Lee, 1972), and may mean that the same resonating/radiating system, possibly on the hindwings, is utilized in both contexts.

The most distinctive component of the tiger beetle’s ultrasound-triggered behavior is the swing of the elytra backwards towards the hindwings. Our latency data show that the elytra swing always occurs before clicking begins. We have preliminary data from ablation experiments and stroboscopic photographs that suggest (a) that the elytra swing is necessary for click production and (b) that the hindwings may strike the elytra during a response, sometimes hard enough to deform the hindwings. As our current primary hypothesis for the click production mechanism, elytral impact with the hindwings has the added appeal of an elegantly crude economy: the single, simple motion of the elytra could alter the aerodynamics of hindwing motion (as happens when the cricket’s leg hits its hindwing; May and Hoy, 1990) and at the same time produce loud ultrasonic clicks.

We thank Dr David Pearson for identifying the beetles. Amy Harron assisted with data collection and analysis. Dr Pearson and Dr Winston Bailey made very helpful suggestions on the manuscript. Dr Kevin O’Grady provided statistical advice. This research was supported by research grant 5-R29-DC01382 from the National Institute on Deafness and other Communication Disorders, National Institutes of Health to D.D.Y.