To determine whether the frog embryos were hatching in response to vibration and not to some chemical or visual cue, Warkentin played the recordings back through a device that mechanically shook the clutches. Sure enough, the eggs hatched more often in response to snake-attack recordings than rainstorm recordings.
"They hatch when the snake starts biting the clutch, not before," she says. "It's not because there are snakes in the neighborhood or snakes there looking at it."
Warkentin examined the recordings to see whether she could tell how rain and snake-attack vibrations differed. "Two of the most obvious differences between vibrations caused by rain and vibrations caused by snakes are elements of the temporal pattern," she says. "Snake bites, in general, last longer than raindrops, and the spaces between snake bites are generally longer than the spaces between raindrops."
To test the hypothesis that embryos use these features of vibrations to assess danger, Warkentin clumped the rain recordings together into tightly spaced segments, making them more snake-like. Conversely, she spliced bits of stillness into the snake vibrations, making them more rain-like. When snake-like rain and actual rain recordings were played to a clutch of red-eyed tree frog eggs, the eggs hatched more often to the snake-like rain recordings. When the rain-like snake recordings and actual snake recordings were played to the clutch, the embryos hatched less often in response to rain-like snake recordings. Both instances, therefore, provided more evidence that temporal patterns serve as an important cue for the frogs.
Warkentin created similar temporal patterns out of computer-generated nonspecific sound, or "white noise." Again, the snake-like pattern of white noise caused more eggs to hatch prematurely than did the rain-like white noise. "These experiments don't rule o