It has been appreciated for a while that fruit flies interact with each other, but studying these interactions is quite difficult. In fact, this is a general problem in understanding animal behavior – the measurements are incredibly tedious to make and to quantify. In a paper published yesterday in Nature Methods (Dankert et al, 2009), Heiko Dankert, Pietro Perona, David Anderson and colleagues now automate the process, using the power of machine vision to analyse social interactions in pairs of fruit flies.
Put two wild-type male fruit flies together, and they will lunge, tussle and threaten each other, actions classified as aggressive (however, not all interactions between male flies are aggressive). Put a male and a female fruit fly together though and, predictably, their interactions will be largely of the courtship variety; the male circles the female, extends his wings in a courtship ‘song’, and sometimes, things work out!
Dankert et al show in their paper that they can take simple videos of fly-pairs, and use new software to accurately identify these stereotypical aggression and courtship actions. This allows them to observe many such fly-pairs and to describe these behaviors quantitatively. What’s more, the software detects expected differences in fly behavior. Male flies in which octopaminergic neurons have been silenced or where the fruitless gene is spliced into a female-specific form are much less aggressive. Male flies in which cholinergic neurons have been feminized now court other males. The software sees these differences.
Importantly, the measurements can be made quickly and without researcher bias or fatigue. For instance, Dankert et al estimate that experiments that would take them about a hundred hours to perform manually can now be completed in a few minutes. This opens up the study of these behaviors to genetic and other screens, and it will be interesting indeed to see how such studies play out.