FE5.2 - Spiders Song (Part 2)

Before we start the show, we want to send a huge thank you to our amazing community on Patreon. Future Ecologies just wouldn't be possible without you, and we are beyond grateful to have your support. We hope it's obvious that every one of our episodes is a pretty considerable effort. Every single patron means more ambitious stories, fair pay for more guest producers, musicians, and other collaborators, and gets us closer to a living wage to do what we love to do —

And I'm Adam.

And this is Future Ecologies. Today, in Spiders Song Part Two, we're taking our seats in the concert hall of life — audience to the grand dance of evolution, with taxonomist, phylogenetic theoretician, and jumping spider devotee Wayne Maddison.

Hi. Good to be back.

In other words, we are jumping in right where we left off.

So do you want to give us a quick recap?

Sure. Jumping spiders are basically like tiny, eight legged, big eyed cats, slash birds of paradise — in that there are bedazzled males that court mates by dancing. And also by singing! In a manner of speaking... they vibrate.

Yeah, go on.

And not only are their species really diverse in shape, and color, they also demonstrate a lot of convergent evolutionary patterns, which are not limited to independently and repeatedly developing color vision, ever more complex courtship rituals, a bunch of them have become ant-like, and there's something going on with their Y chromosomes.

Yeah, mostly. The Y chromosome thing is actually just the one genus Habronattus, not all jumping spiders. But it'll be important later on, I promise. Where we left off in Part One, Wayne was overcome by his sense of awe — that evolution isn't just an endless chaos of diversity. It seems to cohere around certain patterns, motifs, melodies, themes and variations. It seemed to him like the grandest possible symphony. If only he could hear it.

And at first, I didn't know what to do with that. But then I thought "Oh! I'm a computer programmer. I do visualizations of change on phylogenetic trees. Why don't I program a sonification of change on trees?"

I'm assuming what a visualization is to our eyes, a sonification would be to our ears.

Yeah. sonification is like transmogrifying data into sound. In the same way that you might turn that same data into a graph. Sonification is the auditory equivalent.

So last episode, we were figuratively talking about how evolution is a form of music. And now you're talking about literally making evolutionary patterns into music.

Yeah, exactly. So, this practice of sonification has been used to explore and communicate climate data, X-ray astrophotography, prime numbers, and even sequences of DNA itself. But what Wayne is talking about here is sonifying a phylogeny — an entire family tree of many organisms.

A phylogenetic tree is a statement about the history of lineages in the past. And we can't actually go back in a time machine and see those lineages, so we have to reconstruct it. And we can reconstruct it with lots of data, occasionally through fossils. But mostly nowadays, we use genetic data to reconstruct these trees. And it's pretty clear, we're doing a pretty good job of it, because we've got so

So to draw a phylogenetic tree, they have to gather specimens, sample their DNA, and assess them for different characteristics, like which ones have Y chromosomes. Then they use some of the statistical tools that Wayne developed to create an estimate of who branched off from who, and what the characteristics of those ancestors were most likely to be.

So like, if a scientist took you and me, they could cast back and figure out who our most recent common ancestor was and what traits they might have had — based on what you know about us, and maybe some fossils.

And some DNA.

Yeah and a computer program. Okay.

Yeah. So so after they've done that, they have a sequence of all these different lineages, starting from a common root, and then branching and changing through time.

But what would it sound like? Would we hear harmonies would we hear melodies clearly, and so forth? I didn't know.

And as far as I could tell, although this world of data sonification is growing really rapidly, the sonification of phylogeny is unprecedented. Wayne's experiment would be a world first.

I wanted this to have some basis of reality. So I started with a real dataset of Habronattus.

Habronattus, also known as the paradise jumping spiders, most of which are native to North America. And the characteristics examined by that dataset were the various sex chromosomes...

and the issues of the the chiasma localization.

... that is, that is not a term that I am familiar with.

Okay, bear with me for one last piece of cellular biology. If we must. Remember that you've got half of your chromosomes from each of your parents, right?

Yes.

So well, most of the time, the chromosomes from both contributors are paired up, but separate. But during meiosis, the moment at which sperm or eggs are being produced, the DNA from each pair of chromosomes is shuffled together, swapping the copies of genes from either parent. That's the actual moment of genetic recombination that gives you variations.

Yeah, no, no, I'm, I am still with you.

So the chiasma is the crossover point along the leg of the chromosome, where that swap takes place.

Got it! So if you're picturing these cute little X chromosomes with their little dancing legs, right, four legs, it's like, where is that spot where they cross over

And swap.

and swap their information. Yeah, okay.

Wayne had data that included the physical measurements of where the chiasma was located for each of these species of Habronattus jumping spiders. It might be closer to the middle of the chromosome, or closer to the end.

So we were looking for a correlation between where the chiasmata occurred along the chromosome and the evolution of the Y chromosome. And at first glance, you might think "Well, why should those even be connected? It's not as if you needed the chiasmata in a place to generate the Y." So they seemed like two different aspects of the chromosomes.

And in general, correlations like these are exactly what evolutionary biologists are looking for — puzzling out why when one feature is like this, another feature tends to be like that. So Wayne decided to sonify this family tree of Habronattus jumping spiders, comparing the location of their chiasmata with the evolution of a new Y chromosome.

So here's how it turned out. First, let's just focus on the speciation events those points where lineages diverge. Every time you hear a tone, that's a spider lineage splitting in two. The next layer has to do with the chiasmata, where they are in the chromosomes. And because where they are in the chromosomes is variable, like it's a continuous variable, you're gonna hear the tone going up and down in different amounts as the chiasmata slide up or down.

I mean, I think I enjoyed that, because I have a love of John Carpenter horror movies from like the 70s, and 80s, and 90s.

And looking back, I can see that there were a few things wrong with it. The first being how it starts slowly, and then gets busier and busier and busier, as if suddenly all sorts of extra things are happening.

Like it just gets exponentially louder and denser, until it suddenly ends — which isn't really the shape of most music that we tend to listen to.

No, not not mostly no.

So why do you think the data sounded like that?

Well, speciation, right? Evolution tends to become more complex over time. All of the phylogenetic trees that I have ever seen begin with a single line, and split and split and split and split and split until you've got an exponential number more species than when you started. So yeah, it makes perfect sense.

But remember that these trees are constructed by calculating back from species that are still around today. So what's missing?

I mean, we're missing all of the spiders that have gone extinct.

Bingo.

Part of the problem with extinct lineages is that we don't see them today. So we don't know exactly how many there are in Habronattus. And there are no known fossils, it's not like we can figure it out that way. But we can get an estimate of how many there likely would have been. And so one way to do this is to do a simulation of the dynamics of branching and extinction. And we can sort of populate all those

And if you you know, if you were to simulate those extinct lineages, it raises questions about whether you'd want to be able to hear the difference between the real and the imaginary ones. And in the end, with all of the various branches, you still have to deal with a lot of overlapping sound.

The second thing that's wrong, well, there's probably more than one here. But the second thing that's wrong is that you're not able to really hear each of the voices and the melody that it might be playing, because I'm using the same set of notes all through the whole tree. And that what I really needed to have done was somehow distinguish all these voices so that you could hear them separately. So it was almost like I should have said, okay, at the base of the tree at the

And so, four years ago, that's basically where the story would have ended — with a beautiful metaphor, and a not quite as beautiful sonification. And I wasn't satisfied with that. "I... I was wondering... so I asked Wayne, if I could take my own spin at it. "And sort of try to take it to the next step as part of this project."

Sure, I think that could be fun.

And so I tried. But after a few very enthusiastic but ultimately false starts, I too realized that this was a way bigger project than I had anticipated. Not least because at the time I, I didn't really know anything about making music. But it was this project that was my motivation to learn. And even while this project was on the backburner, I fell in love with learning the patterns of music, and with the principles of electronic synthesis. I fell in love with making music just for its own sake.

I enjoy listening to the music you make.

Thank you. You know, looking back, I would say that this was one of my Divide Creek moments. Like this story, put me on a path. And I think I'll be on it for the rest of my life.

I know that feeling. Yeah.

But the other part was that in order to make it happen, I needed help. In fact, I needed a whole team.

Mendel that's called a band.

Well, allow me to introduce Duncan Geere.

Hello, what's up party people?

And Miriam Quick.

The previous slide where we have the phylogenetic tree, does the horizontal axis represent time on a linear scale? Or does it represent some other degree of change,

Duncan and Miriam are information designers, and they're the hosts of a really wonderful podcast that's completely dedicated to data sonification. That's called Loud Numbers. Next,

There must have been a molecular clock.

This is Damian de Vienne, evolutionary biologist at the University of Lyon.

So you have a branch length usually represent the number of mutations that occur along these branch. And then if you have a hypothesis of how fast mutation accumulates, then you can transform that to time,

I did actually end up finding one other precedent for phylogenetic sonification after Wayne's original attempt. It wasn't exactly a piece of music, but more like a proof of concept. Damien was a co author, along with his friend, Henri,

We've done a little batch in pure data, which was sort of a test just to see if it's possible to sonify trees

This is Henri Boutin, acoustic researcher at like that. IRCAM. That proof of concept that I found was really just a side project between him and Damien.

We are friends since a lot of time. We used to do music and things like that. But we've never, we've never worked together. And this was the first opportunity to work together.

And finally, local wizard slash generative music researcher and PhD student, Simon Overstall.

Good morning.

Who joined me in Pacific timezone solidarity whenever we met with our European collaborators.

I need another coffee now.

So what did you do with this incredible team of people?

Well, I think it's probably better if I spare you the prototypes and the meetings and the revisions, I'll just jump straight to what we ended up with. Because just like Wayne's version, I'm going to need to explain what you're about to hear.

Yeah, all of the all of the best music requires extensive exposition, and I am here for it.

Well, in this case, yes.

I'm all ears.

So here we're using the same underlying data as Wayne. We've got these species of Habronattus jumping spiders, we know the location of their chiasmata and whether or not they have Y chromosomes. But the difference between our interpretations starts with how we represent time.

Okay.

The tree itself is the same, and we're not simulating any extinct species. We're just approaching playback in kind of a different way.

But what do you mean by that?

So time still flows from the past to the present. But to avoid that exponential cacophony of all the parallel branches, we decided not to play all the lineages at the same time,

Ah that makes sense. So what did you do instead?

You can kind of think about it as a series of Divide Creeks. We always start at the same place, like the headwaters of the stream, the root of the tree.

The last common ancestor between all of these species

Yeah, exactly. So we follow that one lineage until at some point, it splits in two. Then we follow those two branches. until they both reach the present day. And because the scaling of time by branch length isn't linear, one branch will probably reach its end before the other one. But once they've both finished, we pause and cycle back to the beginning.

So it's basically that they're just two voices at any single point. Got it. Okay.

And each of these trips from the root to the two tips, represents approximately 5 million years of evolution.

Wow. Okay. And how long does it take in like real time.

It kind of depends on which branch are listening to, but a few seconds to tens of seconds.

Got it.

Now, because we're only listening to the branches of this tree one pair at a time, it takes a lot longer to hear the whole thing. But I also think that makes it a lot more musical.

Sure. But what are we actually hearing as we move down the creek? So to speak.

So every time a lineage reaches a point of speciation, where its path might have gone one way or another, it plays a chord. Or more precisely, it plays an arpeggio. Which is like a chord with all the notes spread out. And the notes that are in that arpeggio depend on which daughter lineage our current branch followed, either descending to the right or to the left along the tree.

What do right and left mean in this situation?

So when you're drawing a phylogenetic tree, the order of the branches, and really what's left and what's right... it's all pretty much arbitrary. So this is just a way of having a simple rule about the pitch of the notes that makes each unique branching path, a unique melody.

Okay, yeah, left, right, one way, the other way.

One way, the other way.

And so each unique species plays out as a unique series of notes.

Yeah. Yeah, they all start in the same place, but eventually find themselves somewhere different.

So what is the rule? What are the actual notes in that melody? What do they mean?

Well, the chord that you'll hear the arpeggio is only ever at most four notes. And that's telling the story of four generations. So the great great grandmother note is forgotten. And the daughter note is added. Depending on that, quote, unquote, direction of descendants, a daughter note might be either a minor seventh above the pitch of its mother. Or a perfect fifth below.

Okay, so as we go, we forget a little bit about our ancestors. We may not know exactly what those species were, or what their names were or what their dances were like.

Yeah. But we do still have some sense of where we came from.

Yeah.

Who we came from.

Yeah.

Also, it's important that I point out that the arpeggio isn't in order of oldest to youngest, it's just in note order, either going up or down.

And if it's descending or ascending, then it just gets put in its place. Okay.

And to keep things musical, the notes will wrap to a four octave range.

Okay.

But the melody isn't actually the important part.

That's what drummers tell me.

It's true. So in this case, it's really just describing the shape of the tree. What we're trying to hear is a correlation in the data, a connection between the evolution of a Y chromosome and the location of the chiasmata — these two seemingly unconnected aspects of jumping spider biology.

Oh, yeah. Okay.

So what I want you to pay attention to is the envelope of each note. That is, the shape of the sound — either short, and plucky or long and sustained.

Right. Waaauwww. And what does the envelope tell us?

That's the position of the chiasmata, those crossover points on the chromosomes. The closer the chiasma gets to the tip of the chromosome, the pluckier the note.

Got it.

And the evolution of a new Y chromosome is signaled by a few things as they arrive along the branch. What you'll first hear is a triangle ringing out. Then when you hear the arpeggio, you'll notice that the direction will change from ascending to descending. So what I want you to listen for is how often pluck your notes are arranged in a descending arpeggio. Remember the sound of a triangle is your cue that a Y chromosome has arrived.

Okay. Will there be a quiz at the end?

No, you can just enjoy yourself.

Okay.

Anyhow, that's, that's the main correlation that we were trying to listen for. But we didn't stop there. Next we took Wayne's suggestion that the voices really ought to evolve more than just in terms of melody, but also, timbre,

Timbre, so like the character of the sound. So do they split into like the strings and the woodwinds? And so on and so forth?

Well, not exactly.

So what I gather from all of that... is that things are changing.

That's the case. The more the spiders mutate, the more they sound like different instruments. And this is actually like a way of describing the evolutionary distance along a branch. And one thing I find interesting is how suddenly these changes can sometimes happen. It could be an artifact of how we've processed the data. But it also seems that evolution can be a lot less gradual than we usually expect.

Yeah, that reminds me of a concept that we call punctuated equilibrium, which is just that, like in evolution, things sort of can stay very stable for quite some time. And then suddenly, there's a bunch of fairly large changes, right? The environment shifted dramatically in some way or there was a development of some kind of mutation. And everything happens all at once.

Exactly, yeah, it kind of comes out of nowhere. And another kind of subtle thing you might notice is that where the voices are positioned in stereo, mimics their location on the tree. So you'll hear them moving around your head, as they follow their branches, getting a little quieter as they go out towards the tips.

Of course, of course, stereo can be part of this, I didn't think of that.

So I hope you're wearing headphones.

Never listen to Future Ecologies without your headphones.

We really appreciate it. Lastly, to mark time, between each cycle of dividing creeks, before we return to the root of the tree, you'll hear a short clip of one of our spider friends singing. We processed that through a synthesizer that models the physics of a plucked string, providing a kind of drone for the entire piece — as though the spiders themselves are strumming.

So this is the spider playing a guitar, so to speak. Wow.

That's majestic. I... I love that.

And now, Spiders Song, take two, in its entirety.

That is really cool. It's... it's really beautiful. That's not at all what I would have expected. The sense of how how rich are the spiders in these lineages comes across, you know, it's it's not... it's... the multi dimensionality of it becomes clear, right of all of this.

I feel like I want a whole collection of different phylogenies sonified like this, and just put them on and let my brain simmer

The thing that I'm trying to locate is whether or not the pings... how they're connected with one another, and they're occasional enough that it's hard to find them. Right, that could just be because the data is not showing it clearly.

So is that a symphony? No. And I think that's okay. This isn't supposed to be the way we listen to phylogeny, to the music of evolution. It's just a few ideas for how we could. And if you want to build on this one, I'm making the whole patch open-source. That'll be up on our website,

futureecologies.net I can't wait to hear some spider remixes, or even some other phylogenies put through this system.

Me neither. But, you know, maybe first, it's worth asking, what's the point of this whole exercise?

I can totally be the person that asked that, Mendel. What is the point?

So I guess I just want to make a distinction that there are really two big types of sonification. And across all of them, the goal is always to get the data to speak. But in my case, the key is that I already knew the story that I wanted to tell. And I wanted it to sound good, right? I wanted it to be at least a little musical.

You wanted to tell a story and you wanted it to sound good, which is why you make a podcast, presumably.

That's why we're here! And the data were going to be there no matter what, right? And I realized that being able to really hear them, to hear the meaning and the patterns was so dependent on how I... how I tuned the whole system towards those. But if I were actually trying to do science, to discover something new, it would have been a completely different exercise. And that's really the difference between the explanatory and the exploratory.

So yeah, it sounds like, you know, to me that you were interested in the challenge. You were interested in developing your skills musically. And you were interested in telling this really interesting story.

Yeah.

But, you know, devil's advocate over here. Does this have any scientific utility, like, could data sonification for phylogeny be useful?

For a lot of things we are still in an exploratory mode, and we don't have the hypotheses yet there. And, you know, maybe it'll turn out that you somehow tweak this so that it handles genomes in a particular way, and it's something to do with, I don't know the shapes of proteins or something like that. And you start playing it, and people start noticing patterns from the way it sounds that then turn

This is, I don't know if it's tangent, but I studied experimental film as my undergrad. I'm a film school dropout. And I loved experimental film, not because I would stare at it really hard. And think about it really hard. And try to derive the meaning from all of the kind of madness up there on the screen. No, I would sit there watching those films, often late at night, in a lecture hall with very few

You're welcome. And yeah, the science is just one part of it. I felt like my job was to honor the beauty of these little spiders.

They are quite beautiful. I guess that raises the question like these spiders have presumably evolved all of these things to appeal to one another. Why do you think that they're so captivating to us as well?

At some level, it's a coincidence, right? Like, it's just happenstance that female jumping spiders seem to respond to the same sort of things that we do, right, like flashy colors, and interesting vibrations, just like us. So male jumping spiders have evolved to be dazzling — dazzling in ways that appeal to both of us. And I think, over evolutionary time, jumping spiders are literally being shaped by you might say, their own attention to beauty.

You know, science is typically defined by the very rigorous style of testing that we do. But there's the other half of that, which is the generation of ideas that we then subsequently test. And that generation of ideas doesn't have to come in any rigorous way it can come from anything. And an attention to beauty, that's jostling the way we look at the world. It's giving us surprises, it's helping us to notice things

For all of us, scientists, musicians, and maybe even jumping spiders — our sense of beauty is part of our intrinsic motivation. Each of us, in our own way, witnesses the world, and responds to it. Because there is no such thing as beauty without an audience. This series of Future Ecologies was produced by me, Mendel Skulski, but not without help from so many others. Thanks to my amazing sonification collaborators, Damien de Vienne, Miriam Quick, Duncan Geere, Simon Overstall, and Henri