• Hofstede, Hannah
  • Ratcliffe, John M.
  • Fullard, James H.
  • University of Toronto
  • University of Southern Denmark


We investigated the relationship between predator detection threshold and antipredator behaviour in noctuoid moths. Moths with ears sensitive to the echolocation calls of insectivorous bats use avoidance manoeuvres in flight to evade these predators. Earless moths generally fly less than eared species as a primary defence against predation by bats. For eared moths, however, there is interspecific variation in auditory sensitivity. At the species level, and when controlling for shared evolutionary history, nocturnal flight time and auditory sensitivity were positively correlated in moths, a relationship that most likely reflects selection pressure from aerial-hawking bats. We suggest that species-specific differences in the detection of predator cues are important but often overlooked factors in the evolution and maintenance of antipredator behaviour.


We collected moths and conducted our experiments from June to August 2006 and 2007 at the Queen's University Biological Station (QUBS), Chaffey's Lock, Ontario, Canada. Moths were decapitated (as in Fullard & Dawson 1999 and Surlykke et al. 1999), which renders the moth quiescent and removes the possibility of a shift in tympanal response due to acoustic stimulation (Windmill et al. 2006), and fixed ventral side up to modelling clay using metal struts with the wings spread to reveal the ears. Following thoracic dissection, a stainless steel electrode was hooked onto the exposed auditory nerve and a reference electrode inserted into the abdomen. Neural activity was amplified (Grass P15 AC amplifier) and displayed online using a data acquisition board and oscilloscope emulating software (PicoScope v. 5.10.7, Pico Technology). Sound pulses were generated by a Matlab application and delivered to the moth preparation via a high-speed data acquisition board (National Instruments, BNC 2110), amplifier (Avisoft Bioacoustics, model 70101) and speaker (Technics leaf tweeter, EAS 10TH400B), positioned 30 cm from the moth's ear. Pulses were 10 ms in duration (plus a 1 ms rise/fall time) and produced every 500 ms at a sampling rate of 500 kHz.

We generated audiograms (threshold–response curves) by broadcasting randomized frequencies from 5 to 100 kHz at 5 kHz increments and increasing the amplitude for each frequency until pulses consistently elicited one or two action potentials from the auditory nerve. The voltage produced by the amplifier at this intensity was converted to sound intensity (dB peak equivalent SPL) by calibrating the speaker to continual tones with a Brüel and Kjær 6.35 mm condenser microphone (type 4135) and measuring amplifier (type 2610). We took two measurements from each audiogram: (i) best threshold (lowest intensity in dB to elicit a neural response) and (ii) overall sensitivity (area between the horizontal line set at 90 dB and the response curve (pascals×kHz), see Fullard 1982 for details; figure 1). Table 1 provides a list of species included in this study with data values and the sources of data. Data were the means for each species. Log-transformed data for overall sensitivity explained a significant amount of variation in best threshold (r2=0.49, F1,15=14.46, p<0.002). We feel the former is a more ecologically relevant indicator of sensitivity to the echolocation calls of sympatric bats because it incorporates data on thresholds for all frequencies (Fullard 1982). Therefore, we consider only overall sensitivity here.