It's an everyday scene, which only happens at night. A hungry bat pins its acoustic gaze on a flitting moth, engages its high-precision echolocation system and closes in for the kill. But something goes wrong at the last moment. The bat unexpectedly pulls back within millimetres of its target, and the tiger moth goes free. What triggered the bat to abort its attack? There are several possible explanations according to Jesse Barber. Tiger moths react to bat attacks with trains of ultrasound clicks, which could possibly jam the bat's echolocation cries, startle the airborne attacker into retreat or even warn the hungry mammal that it could be in for a foul tasting surprise. But each type of stalling-call would need very different characteristics; warning calls would need to be delivered early in the attack for the bat to get the message, while jamming signals would need to be targeted late in the bat's approach, milliseconds before the bat strikes. Curious to know when tiger moths respond to bat attacks, Jesse Barber and Bill Conner began recording ultrasound responses from hundreds of tiger moth species as they reacted to an incoming bat (p. 2637).

With few tiger moth species available in North Carolina, Conner directed Barber to the fringes of the Ecuadorian rainforest on the slopes of the Andes,where thousands of species flourish. Setting up a temporary lab at a ranch near Santo Domingo de los Colorados, Barber trekked out into the forest every night to attract tiger moths with a UV lamp. He was inundated. Thousands of insects swamped the lamp. Collecting tiger moths in convenient plastic shot-glasses, Barber photographed each insect, for later identification,before placing them in a sound proof box and playing a 2.1 s ultrasonic bat attack while recording the insect's responding clicks.

But before Barber analysed the recordings, he made a short visit to Becky Simmons at the Smithsonian Institute to identify the 130 tiger moth species he'd captured in South America. After days of rooting through dusty specimen drawers, the pair had identified about 80% of the insects, but many of the moths were new to science.

Returning to Conner's Wake Forest lab, Barber began the colossal task of scrutinising the 700 ultrasonic recordings from 36 species. First Barber timed the onset of the insect's reply relative to the incoming bat's echolocation signal, to see if he could distinguish late jamming signals from early warning messages. He also calculated the percentage of time during a 100 ms window when the insect was clicking, to see what the probability was that the clicks could jam the bat's guidance system.

After months of painstaking analysis, Barber was amazed to see that on average, all of the moths began clicking early in the bat's attack, 960 ms before the bat struck, too early for the clicks to jam the approaching bat's echolocation signals. And when he plotted the percentage of time that each insect clicked during a 100 ms window against the point in the bat's attack when the insect began responding, he realised that clicks probably couldn't jam the incoming mammal's guidance system.

Barber emphasises that his recordings don't rule out that the possibility that the moths have evolved effective jamming, but points out that knowing when the moth begins reacting to the bat attack isn't sufficient to distinguish a warning from a jamming signal.

Barber, J. R. and Conner, W. E. (
). Tiger moth responses to a simulated bat attack: timing and duty cycle.
J. Exp. Biol.