Tiger Moth Jams Bat Sonar

See allHide authors and affiliations

Science  17 Jul 2009:
Vol. 325, Issue 5938, pp. 325-327
DOI: 10.1126/science.1174096


In response to sonar-guided attacking bats, some tiger moths make ultrasonic clicks of their own. The lepidopteran sounds have previously been shown to alert bats to some moths’ toxic chemistry and also to startle bats unaccustomed to sonic prey. The moth sounds could also interfere with, or “jam,” bat sonar, but evidence for such jamming has been inconclusive. Using ultrasonic recording and high-speed infrared videography of bat-moth interactions, we show that the palatable tiger moth Bertholdia trigona defends against attacking big brown bats (Eptesicus fuscus) using ultrasonic clicks that jam bat sonar. Sonar jamming extends the defensive repertoire available to prey in the long-standing evolutionary arms race between bats and insects.

The ability to pinpoint airborne insects in darkness, by echolocation (1), allowed bats to master nocturnal insectivory (2) and set the stage for the evolution of defensive countermeasures by insect prey (3). Some insects gained ears (4) and evasive maneuvering (5). Tiger moths (Lepidoptera: Arctiidae) developed the ability to click ultrasonically in response to attacking bats (6, 7). Decades of research on moth click defenses have led to three, not mutually exclusive, hypotheses regarding their function—startle (8), acoustic aposematism (“warning”) (814), and sonar interference (“jamming”) (1517). When ultrasonic clicks are paired with unpalatable prey, bats learn to perceive clicks as a warning of unprofitability (8, 1014). Moth clicks also startle inexperienced bats or bats that have not heard clicks for multiple days (8). Because bats habituate to startle quickly, its effectiveness as a defense requires clicking moths to be rare. This situation does not appear typical in nature (12).

Finally, moth clicks may disrupt the sonar of an attacking bat (1517). Clicks might diminish a bat’s acuity in determining target distance (1719) or feign echoes from objects that do not exist (15). However, evidence that moth clicks can disrupt bat attacks by jamming sonar is lacking. One recent study found that moth clicks had no discernible effect on attacking bats unless clicks were paired with defensive chemistry (13). All previous studies, however, tested moths with relatively low duty cycles, or sound production per unit time (20). High-duty-cycle moth clicks, such as those of Bertholdia trigona (Fig. 1 and table S2), are more likely to jam bats (17, 18), including the big brown bat (Eptesicus fuscus), which co-occurs with B. trigona in much of Central America and northward to Colorado, USA.

Fig. 1

Jamming acoustic complex. (A) Scanning electron micrograph of B. trigona sound-producing organ (tymbal). Some scales were removed for clarity. (B) Example oscillogram (time × voltage) and (C) spectrogram showing a simulated E. fuscus echo, reflected off of B. trigona at 1 m (left), and two overlapping B. trigona click modulation cycles (right). Echo and clicks are presented at average estimated intensities at the bat’s location 1.0 m from the moth (21).

To determine whether the clicks of B. trigona startle, warn, or jam bats, we pitted E. fuscus against B. trigona in a flight room equipped with high-speed infrared video cameras and an ultrasonic microphone (21). The pattern of a naïve bat’s success in catching clicking moths, when measured over several days, should differ for each of the three proposed moth click functions (Fig. 2A) (13). For a warning sound, bats initially capture and drop distasteful moths, then learn to abort attacks on hearing clicks (8, 13, 22). Alternatively, a bat could be initially startled by the clicks but would quickly habituate to the sounds (8, 23). A jamming defense, however, would immediately deter attacking bats, and the effect of the defense would persist with time (Fig. 2A).

Fig. 2

Predicted and observed success rates for E. fuscus attacking moths using various defensive strategies. (A) Predicted bat attack success over successive nights for three proposed moth click functions: startle, aposematism, and jamming. (B) Observed bat attack success rates for three E. fuscus (J1, circles; J2, triangles; A1, squares) on B. trigona (closed symbols) and noctuid controls (open symbols). The solid line shows mean values for attacks on B. trigona. (C) Success rates of bats attacking intact (sound-producing) and silenced B. trigona. Means ± SEM. Some values were offset slightly for display.

We trained three naïve juvenile E. fuscus (bats J1, J2, and J3) and one adult E. fuscus (bat A1) to capture tethered moths in a sound-attenuated flight room. Naïve bats allow us to control prior experience, whereas an adult bat’s prior experience catching insects may allow it to better overcome moth defense. On each of nine consecutive nights, 16 tethered moths—4 B. trigona; 4 silent, palatable, size-matched noctuid novelty controls; and 8 silent, palatable wax moths (female Galleria mellonella, used in pre-experiment bat training)—were presented to each bat individually in a random order. Bats were allowed up to 1 min or five approaches for each moth. We presented unaltered B. trigona to the bats on nights one through seven and B. trigona lacking the ability to click (from tymbal ablation) on nights eight and nine. Seventy-eight percent of the B. trigona (87 of 112) responded acoustically to bat sonar. The few intact B. trigona that did not produce sound in response to bat attack were excluded from the analysis. We used contact with the moths as our measure of attack success. Palatability of B. trigona to each bat was taken as the proportion of captured moths eaten rather than dropped. We conducted feeding trials with silenced moths and four additional stationary bats to further assess the palatability of B. trigona. Finally, we analyzed echolocation recordings to measure bat behavioral response to moth clicks (21).

In stationary feeding trials with silenced moths, three bats regularly ate B. trigona (J4, 92%, n = 12; J5, 92%, n = 12; J6, 83%, n = 6), and one did not (J7, 0%; n = 8). Similarly, of the bats capturing moths in flight, three regularly ate B. trigona (J1, 94%, n = 16; J2, 50%, n = 14; A1, 89%, n = 18), and one did not (J3, 0%, n =12). The observed differences between individuals may reflect a greater reluctance of some individuals to incorporate new items into their diet, as has been reported for some birds (24). B. trigona appears palatable; the six bats ate the moths with no detectable ill effects. Because warning and jamming are not mutually exclusive defense functions (25), our ability to detect a jamming effect is diminished for the bat that rejected B. trigona—any moth avoidance could be caused by jamming, warning, or both. Thus, to unambiguously test the hypothesis of sonar jamming, we focus our analysis on the three bats that ate B. trigona in flight (J1, J2, and A1).

Each of the three bats contacted control moths over 400% more often than clicking B. trigona (Table 1 and movie S1). In contrast, bats contacted 100% of B. trigona when the moths’ tymbals were ablated (Fig. 2C and movie S2). This confirms that the bats’ poor performance on B. trigona was caused by the moth clicks. Rates of contact with sound-producing moths did not change during the experiment (Fig. 2B) [Friedman nonparametric analysis of variance (ANOVA), χ2 = 2.63, df = 6, P = 0.85]. Thus, prior learning was not required for the defense to be effective, nor did the bats habituate to the clicks. Bat contact rates did not improve on succeeding attacks on the same moth—attacks that often occurred within seconds of each other (table S1) (Friedman nonparametric ANOVA, χ2 = 2.55, df = 4, P = 0.64). This renders startle unlikely because nearly all mammals, including bats, habituate to startle (8, 23). Echolocation recordings (Fig. 3) showed bats persisting in attack attempts after onset of moth clicks to varying degrees throughout the experiment (fig. S1A). In contrast, bats in a previous study quickly learned to abort attacks on the noxious Cycnia tenera (fig. S1B) (22). This further suggests that the bats in our study did not learn to take the moth clicks as a warning—the bats continued attacks despite hearing the sounds. It also indicates that moth clicks are not generally aversive to bats [supporting online material (SOM) text]. Together, we take the above results as strong evidence for a jamming function of the moth sound. The defense was effective immediately and persistently on juvenile and adult E. fuscus (Fig. 2B) that frequently tried to capture the clicking moths (fig. S1A) but had much difficulty doing so.

Table 1

Contact rates of big brown bats on B. trigona and noctuid control moths.

View this table:
Fig. 3

Example echolocation attack sequence. (A) Spectrogram showing a big brown bat attacking a clicking B. trigona. (B) Pulse-interval graph of echolocation attack. Each circle connected by solid lines indicates the time of a sonar pulse and the duration of the following pulse interval. Open triangles indicate moth click half modulation cycles. The dotted line shows an attack on a control moth for reference.

We also observed atypical echolocation behavior in response to B. trigona clicks. Each bat attacked all control moths by progressing through the approach, track, and terminal phases [e.g., (26, 27)]. However, in about a third (J1, 21%, n = 19; J2, 40%, n = 15; A1, 28%, n = 25) of attacks on B. trigona, the bat reversed the attack phase from track to approach or terminal to track, and then continued the attack (Fig. 3). This behavior did not diminish with experience (Friedman nonparametric ANOVA, χ2 = 4.5, df = 6, P = 0.61), as would be expected with startle. It appears to be a direct response of the bats to the sonar interference caused by the moth clicks.

The ultrasonic clicks of B. trigona appear well-suited for jamming echolocation (SOM text). Under all proposed jamming mechanisms, high-duty-cycle sounds made by highly developed tymbals (Fig. 1) should better jam echolocation (15, 1719). Empirical evidence now supports this prediction; the high-duty-cycle sounds of B. trigona jammed bats, whereas the low-duty-cycle sounds of Euchaetes egle were ineffective jamming signals under nearly identical conditions (13). The use of tiger moth sound for aposematic signaling requires only simple tymbal structure and appears widespread in the Arctiidae (7, 1013). Thus, acoustic aposematism may have been a stepping-stone in the evolutionary development of a jamming signal, a unique acoustic countermeasure in the long-standing arms race between bats and their insect prey.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 and S2

Tables S1 and S2


Movies S1 and S2

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank F. Insana and the staff of the Southwestern Research Station for assistance collecting B. trigona; M. Cullen for providing bat care and training; N. Hristov and J. Muday for technical assistance; and B. Chadwell for software. M. Conner, J.-N. Jasmin, T. Eisner, and the Wake Forest University ecology group provided editorial comments. Funding for research was provided by National Science Foundation Grant IOB-0615164 to W.E.C. and a W.F.U. Dean’s Fellowship to A.J.C.
View Abstract

Navigate This Article