Explaining Fly Song

I spent the last four years studying fly song. When I tell this, many people are surprised and ask: flies sing? And: yes, they do! And I explain how and why flies sing in this video. (You may need to use Google Chrome Browser to view this video.)

Song is part of fly courtship. Male flies produce a song by extending and vibrating one of their wings. And females ears’ are specialized to hearing song.

Interestingly, fly song is structured and males sing different types of songs: sine and pulse songs make up unique patterns. Sine song is a low humming noise and pulse song sounds more like beats on a drum. Females pick up differences and can tell if they like a male from hearing his song.

Flies are tiny and people often ask me, if fly song is audible for humans and yes it is: you could hear fly song, if flies were courting in your ear. The reason we don’t hear fly song every day, is not because it’s not audible, but because it’s not very loud. Imagine: male fruit flies produce song by shaking their tiny wing—even when they shake hard, they can’t produce loud sound.

So one important part of my work was to record elusive fly song, so that I can study it. To do that I worked with brilliant engineers, who helped me build a song box, with a very sensitive microphone, that’s just perfect to pick up flies singing.

Now you know that flies sing, and that we can even hear their song. Probably you want to hear it now, don’t you? You can: check out this video of flies courting and singing!



Female fly choice

Imagine, you are in a white room and it is quiet. And there is this guy with you and he is frantically running around you while he is strangely quiet. He is freaking you out, you want him to leave. But then you start to hear singing. It seems to come from this guy and it sounds beautiful. Someone who can sing that beautifully, can’t be that bad. Actually he is kind of cute in his frantic way. So you allow him to get closer, you allow him to touch you and end up having sex with him. It is a bit odd that he continues singing all along, but you don’t mind, you like his song.

This is the actual setting that a female fly finds herself in when I am doing my experiments. She is locked in a white room, which we call courtship chamber, together with a muted male, that she finds unattractive as she runs away from him and kicks him when he comes too close.  And then, at one point, I start playing back fly courtship song to them and the female’s behaviour changes. She slows down and allows the male to come close and finally copulates with him. However, females are picky, and do not copulate to any song, no, it needs to be the right one.

And I find that the right song needs to have the right song amplitude structure, that is the right way to vary in volume. In specific females like song amplitude structure typical for their own species, which you can see from these data that show what percentage of females copulate to their own species song and to a different species song:


Not available

Strikingly females from both species, copulate more when hearing a song with their own species song amplitude structure, however they don’t like song with another species song amplitude structure. Actually the song is rendered as unattractive as silence, which when you think back to the beginning of this post must be really unattractive.

What’s your favourite song, fly?

When I want to know what’s your favourite song, I just ask. It’s as simple as that. But how do you ask a fly what’s their favourite song?

Clearly, flies cannot tell me what song they like as a person could, but you can tell what songs flies like from their behaviour. This is because, when flies hear a song they like, they change their behaviour, they actually start to dance and sing along – no kidding. Well it’s a bit more complicated than that as only male flies actually sing along, but both males and females start dancing, or as scientists would say, they increase their locomotor activity and chase one another. But really, moving and chasing is what a dance is, so I will call it dancing.

Flies don’t dance and sing to any song, no, they are picky. For example they don’t sing to Travis’ Sing, sing sing despite the clear encouragement to sing.

Flies only dance to songs that meet certain criteria: they need to be at the right frequency and come in the right rythm. The right frequency and ryhthm are determined by the species the fly belongs to, as fly species sing differently and like their species’ songs most (see my post from the 27. April ’16).

So how do we find out what songs flies like? We need flies, a speaker, fly song for playback, and fly chambers to prevent flies from escaping. You then place flies in the chambers and start playback. If flies like the song they’ll start dancing and females become more willing to copulate (note it’s female flies, not humans).

I am interested to find out whether flies care about variations in volume over time. So I designed fly songs with variations that are typical for three different species: melanogaster, simulans and yakuba. For that I calculated the mean gain for each species, generated an envelope with that gain and masked songs of the three species, as you can see here:


So which of the songs do flies like? I will let you know in the next post.

Very variable volume – Decoding song amplitude structure in flies

„Louder!“ The radio plays that film music that gives me the creeps: Lux Aeterna. It is an orchestra that starts off quiet and then grows into a loud storm, while in the film banality grows into horror. However it’s not the film, but the music that makes me feel uncanny. Now imagine the violins kicked in loudly and continued loud – that’d feel different, maybe even boring, wouldn’t it? Variations in volume are frequently used in music, which is why they have technical terms as ‘crescendo’ and ‘decrescendo’ describing increasing and decreasing volume respectively.
However I am not a musician, but a scientist and want to tell you about crescendos and decrescendos in the song of fruit flies. Yes, flies sing! Check out this video to see how. You can see increasing and decreasing volume in this image of a fly song recording:


Remarkably, other flies make sense of these variations in song volume: flies like volume variations typical for their species better than others. Finding what volume variations are typical for a fly species is not trivial, but doable. I illustrate it in this flowchart:


First I define song amplitude structure (SAS) as volume increase across a local volume peak (circled dots) followed by decrease to a local volume minimum. Then I detect pulse maxima and minima (dots) in courtship song recordings and arrange the pulse peaks relative to SAS peaks. Finally I present SAS data as a heat map. You can think of this procedure as decoding a temporal structure encoded in fly song. This uncovers a diamond-like structure that differs between species, as you can see here:


Flies imaged by Nicolas Gompel.

But how do flies produce these structures? And how can you test what structures flies like? This post is too long already, so I’ll try to answer these questions in later posts.

Stats Visualisation

While looking up Cohen’s d, an effect size for significance testing, often used by psychologist, I came across a website that does a great job in visualising standard statistics: rpsychologist.com

My personal favourites are Interpreting confidence intervals, as it visualises the dynamic nature of sampling over time and Understanding Statistical Power and Significance Testing because of this quote:

“What’s wrong with [null hypothesis significance testing]? Well, among many other things, it does not tell us what we want to know, and we so much want to know what we want to know that, out of desperation, we nevertheless believe that it does!”

– Cohen (1994)

Presentations Do’s and Don’ts.

The university of Stanford has added ‘Mining Massive Datasets‘ as a new course to coursera, a website which offers free online courses.

Being capable of ‘Mining Massive Datasets’ is undoubtedly a handy skill-set. Also I highly approve of open-access courses; however when I watched their tutorial introductary presentation I bursted out with laughter in disbelief. Have a look yourself:

To me, clearly, Jure Leskovec, Anand Rajaraman and Jeff Ullman, who present the course, could have done with some media counselling:
The background looks as if it was taken from a 90′s computer game. This is a computer science course, yes, but this does not mean the presenters need to be in a virtual server room. A real-life course room would have done the job.
Non of the moderators moves while presenting. Moving is natural, sitting stiffly not.
You should not change the position of moderators after a hard cut. This makes the presentation involuntary comical: Jure disappears and Anand pops up, surprise, surprise. You can make smooth transitions with overarching sound and close-ups.

Take this as an example for a good presentation; never mind the content:

Jure Leskovec, Anand Rajaraman and Jeff Ullman have obviously made an effort to present their course: they made sure to have good sound and video quality. The shortcoming of their video is not the video quality but the presentation’s quality.

Disco Drosophila

Mark Jones and Keiran Foster from the Physics Department in Oxford helped me building a device which looks like a disco for fruit flies:

The video was made by my husband Christoph out of a recording I made. Keiran helped me by switching on and off the LED’s.

I will use the red and green LED‘s to depolarise genetically modified neurons in fruit flies. This technique is called optogenetics and was developed by Boris Zemelman and Gero Miesenböck. Gero’s name links to a TED talk of him about Optogenetics.

I call the disco device ‘song box’. This is because I will use it to record fly courtship song. In the video you can see a dark spot in the middle of the box - it is a highly sensitive microphone for recording fly courtship song. Male flies sing by extending and vibrating their wing. They do this when they see female flies. However they will not sit and wait on top of the microphone out of good will. I will restrain flies in a small chamber which fits on the microphone.

My first poster price

I won this year’s ‘best poster prize’ at the European Conference for Mathematical and Theoretical Biology (ECMTB). The poster’s title is ‘Decision making models of courtship song production in Drosophila melanogaster‘.


This is a picture of happy me, the poster winner’s diploma and my new friend, rubber plant.

The rubber plant and I befriended in the moment when I started feeling awkward being photographed standing alone in front of a wall holding up a sheet of paper. In that moment I embraced the rubber plant and we became friends…