 It's one month long workshop for zero-division for central energy. Maybe more of a binary application. It's organized by Matthew, who is truly Susan. My sorority of power and peace. So, guys, please, thank you for that. Everybody, welcome to offline. By the way, officially live is actually very soon. Not after us, but after KDP. So, please, apply even if you just want to record in the room of this workshop. It's not a conference. Basically, you come into a set of people's lives. The minimum time is about two weeks. If you need to come into a set of people's lives, only one day, KDP, the rest of the time, you can work with other stuff. If you imagine, I wouldn't say any more time. This should be fine. The speaker is Michael Wittenson, who has been part of the faculty in this Chicago community. It's in Washington. Okay. Thanks. Well, like everybody, I want to thank the organizers for inviting me, especially since I can't really say that I'm a card-carrying member of the olfaction community. But it's great to be at least a temporary member. And I'm learning a lot. It's like to reach in from a fire hose. So I want to talk about the natural behaviors of flies and the role that olfactory behaviors play in certain aspects. And so just as a very brief introduction, very similar to what Bill Hansen said a few days ago, just to give you a little bit of what motivates me as a scientist, especially for this particular problem, I mean, you've probably all seen pictures like this that show that at least from the perspective of, you know, biomass or species diversity, insects are the most successful radiation organisms in the history of life, as Bill mentioned. What Bill didn't mention is sort of what zoologists think insects are so successful. And most would say this, it's flight. Flight evolved exactly four times in the history of life. Each event was associated with a remarkable, a subsequent radiation, most prominently in insects but also in birds, bats, and pterosaurs. Pterosaurs weren't all these, you know, giant scary things. Most were about the size of pigeons and they were extraordinarily diverse before they went extinct. And most of my research career, I've studied flight, you know, basically because it's cool. I think if I was put it a little more scientifically, flight really pushes the envelope of organismal design. So within flight systems, and by the way, this is a high-speed video shot at 6,000 frames per second of two Drosophila that were on a collision course with one another, and here's a photo montage that shows that they can, you know, exhibit a invasive maneuver within about somewhere around one-fifth to one-tenth of the duration of a human eye blink. And what we find within flight systems is the most powerful muscles that have ever existed, the fastest visual systems, arguably, from talks we saw today, the fastest olfactory systems. I mean, the list goes on and on because it just requires specializations to fly. And why fly? Well, the cost of transport is so low that it basically allows you to carry your body mass a greater distance per unit time than any other form of terrestrial locomotion. Now another thing that I think is really fascinating that's come from sort of modern-day neurobiology is we're learning almost on a daily basis as papers come out that the neural architecture of insects is astonishingly conservative. So we can find homologous neurons, homologous circuits, you know, families of neurons in animals as behaviorally different as social honeybees and fruit flies. And I think that conservation underscores behavioral modules that are very broad and very, very ancient. So we can find navigational mechanisms in ants, modern butterflies, dung beetles, and as I'll show you, fruit flies that are all actually using the same homologous circuits. And if we sort of think about that in terms of evolution, we have to go back between 450 to 400 million years ago when this great radiation took place that has given rise to these crown taxa. And so if there's one thing that's in my mind as cool as insects, it's time travel. And so what I really feel like what we try to do is to understand these behavioral modules because by understanding these behavioral modules that are common among insects, we're really getting a view of what this, which I think is one of the most important creatures that ever lived, the early flying insect that forever changed the world, changed ecosystems, stressful ecosystems in just a radical way. We kind of get to know this guy. So it's like going back in the time machine. And to put this in a little bit of an analogy, this is Pi. Does everybody know what Pi is? Okay, it's kind of interesting to look at these words. Pi is Proto-Indo-European. And so what linguists can do by looking at current languages is reconstruct this. These are English translations of Proto-European. They can put together a vocabulary and sometimes even syntax of this ancient language and these crazy societies that actually try to speak Proto-Indo-European. It sounds a little bit like Klingon if you've ever heard Klingon. But it's the same sort of idea that we get this amazing view into this early culture by trying to reconstruct this ancient language. Just as I think we can reconstruct this ancient creature at the very base of insects. So is there a Precambrian toolkit? So yes, indeed there are. And I want to point out that this is one of people's favorite proposals for a fossil that might be very, very similar to the organism that was at the base of the split between protostromes and deuterostromes. It's called Dickinsonia, which my wife, you know, she says, why are you so excited this? Dickinsonia is usually described as like an enigmatic gelatinous brainless little worm. And they say true. But then that indicates that the ancient features are captured by some crown taxa. So it sort of supports, you know, my general hypothesis. So the behavioral modules that I want to talk about have to do with how animals find food and flight plays an important role because many ecologists have discussed these sorts of issues before but search is really a multi-scale phenomenon. But with an insects and other flying animals that scale gets greatly expanded because you can fly. So you have a long distance component, long distance dispersal that can be from continent to continent that literally can be, say, in some birds from near the North Pole to near the South Pole. And then there's sort of often a sort of broad local search, sort of, you know, trying to find, in this case, the right rotting of fruit. And then once you find that, there's often sort of a narrow local search of finding sort of the best yeast spot within the fruit. And again, this is just outlined for fruit flies. But ecologists working on optimal foraging strategies have talked about these multiple scales for some time. But what I think is interesting as a neurobiologist is that different sensory cues are available, you know, at different spatial scales. And the brain has to use those sensory cues in a sort of scale appropriate way. And that will be sort of a major theme. And just to put this in sort of a tutorial sense. So imagine, you know, a hungry fruit fly would first, you know, try to find this nice orchard. Once it finds this orchard, it's going to want to find a tree that has a lot of good rotting fruit under it. It's going to have to pick one of those and, you know, find which outbreak growth of yeast is going to sample at any given time. So I always like to try to give acknowledgment up at front. And I'm very, very happy and proud to have a, you know, kind of diverse laboratory in all ways. And today I'm going to be highlighting the work of three people, Floris van Groegel, who's been in the lab. And for a number of years now, he's now moved back to Washington. Irene Kamen, especially, Kate Leach, who's a new member of the laboratory, but has pioneered some research that I'm just about to talk about that required enormous amount of skills from sort of field martial type skills to modeling skills and so forth. So the first and most part of the talk is going to be about long distance dispersal. And also I have to thank Massimo, who helped me get the money to do the experiments I'm about to talk about as well as Floris and especially Kate. Okay, so a lot of my thinking on this was transformed by a reading of a paper a while ago that was written by Jerry Coyne. Jerry Coyne is a very, very famous population geneticist and was testing a conundrum proposed by Dubjonski in the early part of the last century about why disparate populations of Drosophila species were so genetically similar, even though they were thousands of kilometers apart. And at that time people couldn't imagine that the flies could actually fly that far. So Coyne, in a series of papers with collaborators, did some remarkable experiments in Death Valley National Monument, where they released, I'm kind of simplifying things, but released 60,000 sort of cosmopolitan Drosophila melanogasterin simulans at 6 p.m. one night from Sheep Creek Oasis and had banana traps in two oases of 15 kilometers and about a seven kilometers away. And don't think of like camels and palm trees and like, you know, sexy people on pillows, oases. These are like little tiny fetid swamp oases, okay? But nevertheless, remarkably, by 9 a.m. the next morning, 17 were caught 15 kilometers away in one direction and about orthogonal direction. 17, coincidentally, were caught at Sheep Creek Spring. So these are animals traveling. We did put this in perspective. So roughly 10 kilometers a day. So, you know, you've heard about, this is the great Arctic turn that goes from, you know, the northern hemisphere to the southern hemisphere. It's traveling about 1.4 million body lengths a day. The moderate butterfly into the famous migrant, 2 million body lengths a day. This traveled 10 kilometers between 3 million body lengths for a fruit fly. Now I've been talking about the experience for a while and this whole idea of the Tobonian Toolkit, which is this notion of all these ancient modules that this, you know, proto-insect must have had, it was all sort of, came from a review paper that I wrote that was focused on these experiments. And a lot of people just say, you know, I just don't believe it. I don't believe that these, you know, so we started out to repeat them. Now Death Valley is now a national park. We can't do the experiments there, but we found an experiment halfway in place, halfway between Los Angeles and Las Vegas. And, you know, you can see it's like a, it's a wonderful place where you want to hang out where there are places called Mars, Zizzix, Midway. I mean, it is a really horrible place. But there's this wonderful white patch there. I'm called Coyote Lake and I'll enlarge that. And I've never seen any coyotes there, but actually the type of species that gives this type of desert, it's like characteristic, like you know you're in New York's desert when you see this guy. This is a sun scorpion. But actually, don't worry, because they're not around during the day, they just come out by the thousands at night. But they're not true scorpions, so they don't have a poisonous tail. Instead they have these nice sharp poisonous fangs. But I've been told that the next time we do these experiments, Massimo has volunteered to like spend, you know, keeping these away from our tent at night. So this is enlargement of the view of Coyote Lake, and it's a dry lake bed. So there's no water. And so we have a release site. We have a trap set up in two rings, a half kilometer ring and a two kilometer ring. And this is what it looks like. It's absolutely unworldly. It looks like the opening sequence in Star Wars, effectively. And here's Kate and Francesca and I setting up what we call bionic buckets, which are traps with a carefully engineered surface. We can with a Raspberry Pi computer, infrared lighting, so we can take pictures of the animals that land on the traps. And then some of those animals actually go into the traps where they get caught and we can count them later. So it's a trap, as Admiral Ackbar says. And I don't have time to go into this, but the surface of this thing is really complicated. It's taken many months to get this right so that we can image on top of it, get it traps the flies. And this is why it's great to have a diverse laboratory where none of the male lug heads would know how to use this machine. And this machine was absolutely essential for figuring out how to get this to work. And what we put in this trap is tree top apple juice, half a liter of tree top apple juice, and some champagne yeast that are fermented for exactly two days. That's all that's in these things. Okay, so here's what one of these traps look like. Yeah. I mean, we wanted to be as close as we could to the original experiments and yes, there's less distractions and it's just this big open surface. But we need to do it in other places as well. So the release site, so this is one of the traps a kilometer away. The release site you can see in large there, there's two little objects. I think that if you enlarge, they will kind of take a magnification of that. And okay, I'm sorry, that's not a release site. Actually, what that is, is one of these covered in a black plastic bag. So here's 30,000 flies ready to go. And I know what you're thinking. Oh, we have a weather station, so we get wind direction, temperature, humidity, all these sorts of things. Yes, we count them. So we count them because Kate does advise a really clever, we have these laser cut inserts that we put inside the fly bottles. The flies pupareate on these things. We can put a bunch of these things in one box and all the flies come out and then we can count this with machine vision or manually and so forth. So we know that we're within a couple thousand of 30,000. Okay, let me just get to results. So these are results. So we released about 25,000 flies in this experiment. This was the wind direction at the time of the release. The numbers of flies we caught in the traps in the half kilometer ring and the two kilometer ring and we were just fucking surprised. We didn't think we'd catch anything and we get all these flies and this is just displayed in kind of graphical form. So I want to first talk about the traps that are one kilometer away from the release site because we fluorescently tagged them with fluorescent powder and we don't need to though. There's nothing... Well, no, okay. I don't want to give my punchline away. There are no other... We've never caught a drosophila that wasn't ours. We have caught other things, but that comes later. So this is just the number of flies on the traps with the camera system, data from one trap and remarkably, the data from different traps regardless of wind direction overlaps. So the flies are compensating for wind which is sort of what they're supposed to do with a hitch, I'll show. So we also... We can kind of figure out from the number of flies we get that about half of the... If a fly lands... If we see the fly with the camera, there's about a 50% probability that we will trap that fly within the next second or so. So anyway, I mean that's not really important, but I want to overlay all the data from all the traps and this is pretty remarkable. If I put this in terms of speed, and this is just using the distance directly from the release site to the trap, the first flies that get there must have flown at 6.5 kilometers per hour, minimum, because they might have waited for a while, right? And then even the kind of these slow pokes, they're still traveling fast, and then all the flies, they kind of get through. It's like a shockwave of flies that's done in 36 minutes. So I'll just put this in perspective. Six kilometers per hour is about like a fast brisk walk. Okay? So if you had a pet fly, you could take it for, you know, a walk. Okay. So again, these are the data. So how do we make sense of these data? And so Kate has been doing some modeling. So this is just a very, very simple model, and I know everybody's going to laugh at the plume model, but it doesn't really matter how sophisticated you make the plume model within this, you know, for the results I'm going to describe. So here's the plumes coming from each trap. And then we do a simulation where we release a fly and every time it crosses a plume, there's some probability that it'll detect the odor in the model. And then if it detects the odor, it does the kind of, the classic search, cast the search, we'll show you, and it gets trapped. But the probability that it traps falls off with distance from the source. And we can change all these parameters, but what I'm about to tell you doesn't really make much difference. So here's a case. This is what we call sort of the Death Star. Keep this star racing. A case where the flies are just taken random heading and these would be the results that we would get. And there's a bias towards the upwind traps, which is not what we see. And so, and this summarizes the results of that simulation. So here's another simulation where we sort of bias the flies downwind and this reproduces the sorts of distributions that we see in the real data. Again, you know, we're in the case of sort of doing thousands and thousands of simulations to all sort of parameters, but this is sort of one of the things that emerges. And this already is pretty interesting because if you know this literature, you know, if there's an idea that if you're hungry and you're searching for something that's wind-borne, you should fly upwind because if you detect something, that's where it's coming from. The other notion is no, you should go crosswind because then you're going to intercept plumes more frequently. But what we're finding is that flies are actually going downwind, right? And I should say, to just give you some intuition, what happens if you go randomly, the reason why you don't get caught at downwind traps is because you hit the plumes of upwind traps first and then you make your way up, right? And so whereas if we have this kind of bias, you can get to the downwind traps, of course then you have to work your way against the wind the long way back up, but this would appear to be what the flies are doing. So these are very early days, but we're pretty excited that these experiments even work at all and now we can do much more interesting geometries and so forth. The wind speed is approximately their flight speed of about six kilometers per hour. It varies a lot in the site, so every time we do the experiment, but it's usually between six and 30, but we can't do the experiments when it's like 30 kilometers an hour. No, no, well, they don't take off when it's that windy, right? Yeah. Okay, so I mean, these are all great questions. I just want to get to some, so one of the things assumption that I've been making is that they fly in straight lines. Now, why do I think they fly in straight lines? I mean, there's a number of different arguments. It's hard to fly in a straight line. These aren't flies, these are humans, okay? These are humans walking with burlap sacks over their head on big open fields. And when you do this, you'll walk in circles. It'll drive you crazy because you won't know you're walking in circle, but you actually walk in circle. And it's just, it's kind of an interesting highlight of how important sensory cues are to be able to fly straight. So in the middle of that vast landscape, how to fly straight. Well, we know a little bit about this because even when there's no sun in the sky, there's a pattern of polarized light in the sky that comes from Raleigh scattering and that makes a coherent pattern of light polarization because when the photons from the sun hit the atmosphere, they get reflected and the angle at which they get reflected relative to the direction of path sets the degree of polarization and you do all the geometry and you have this very coherent pattern that's maximally polarized in the equator between the sun and anti-sun axis. And insects, again, Devonian toolkit, they have been looking up in the sky for 500 million years. So they have all of this hardware for detecting polarized light and this special region called the dorsal rim which flies have and Peter Weir in the lab has been working on this doing two photon imaging showing that there is a map of polarized light orientation across the dorsal rim and this lines up very much with beautiful work from Uwe Hamburg's lab in Locus that there is a central map in a region called the central complex. So flies, you know, if they could use this information they could use it to keep a maintain a compass heading and Peter did some experiments years ago with this really fancy device we were able to build within the laboratory which you can see right there which is what we call a magno tether. So a magno tether is a fly glued to a pin. The pin is oriented between the north and south facing surface of a magnet with a camera underneath that can track the angular orientation of the body. So the fly is fixed in space but it can choose its orientation. We can have either a glass lid or various filters like circular polarizing filters and so forth. We can take this outside and see what the fly does to just a natural sky. And so here's a sped up 10 times movie of a 24 minute experiment and you're looking at the sort of the rear end of a male fly and the fly is maintaining an orientation that's very stable within the coordinate system of the arena for three minutes indicated by these great patches. Peter rotates the arena by 90 degrees in world coordinates and the fly at that moment rotates in arena coordinates and so the purpose of that perturbation is easier if I just should have showed you the time series. So this is what the experiment this is what the data looked like in arena coordinates but if you rewrap it to world coordinates this fly was maintaining a westward heading for 24 minutes which based on our new data would take it three kilometers roughly in the same direction. So when the sun of course is available the sun is a really potent signal and we can do the same trick with another sort of flight simulator that Isabelle Geraldo was using in the laboratory that just a little tiny very bright dot that moves that the fly can control and close loop by regulating its wing motion and the fly will maintain a heading relative to that sun and this is not flying towards the sun because the fly will often pick an arbitrary heading and just maintain it. So this is something we call proportional navigation that's a lot more sophisticated than just photo taxes because at one time it might choose east by northeast and another southwest and this is a great proportional navigation is known it's how you can have an air to air missile can hit an aircraft it's how beetles tiger beetles capture their prey dragonflies capture their prey and it's why moths spiral into flames because they're trying to orient to the moon but unfortunately they've picked an object that is not the moon an object that they get close to so again this is a very very ancient behavior within insects and the reason why we're sort of more interested in sun fixation right now is because it's a slightly more robust behavior possibly because we can just reconstruct the stimulus more easily so we can do an experiment where we let a fly pick a heading we stop it for five minutes and let it go again and see if it picks the same heading or does it choose a new heading and we can kind of we analyze this by analyzing the vector strength which is sort of the addition of all the unitary headings over time and we can plot the heading of the first flight against the second height and we plot the heading twice to just emphasize that these are circular distributed data and the correlation is very very strong and if you take sort of the mean absolute value and difference of the data compared to a scrambled distribution by sort of pairing the data from all the different flies you get a very very very highly significant result that the flies remember their heading across five minutes of no flight they remember it off of two hours of no flight but they start to forget it over six hours and since flies have an activity period of about four hours this would mean that if they once they choose a heading they would keep that heading for basically the whole day so I just want to say a little bit because I know this is probably what people are going to ask me questions about this because there's been a lot of fantastic work from Vivek J. Rahman and Gaby Mamon and Tanya Wolfe on the so called compass system of the central complex so there's cells within the brain of the fly in the ellipsoid body that is thought to represent the azimuthal sort of orientation of the fly and this is from just a summary that the Turner Evans and J. Rahman did and so you might ask the question what are these compass cells doing during the sun fixation paradigm and this is what they're doing that the representation, the angular orientation of the sun is being mapped internally in this toroidal neuro pill within the fly so we're pretty excited about sort of chasing down this circuitry and now I want to get back to another topic which is getting back to odors, what are the flies attracted to? So I'm telling you we just put tree top apple juice and yeast and we've done very very simple preference experiments that if you put apple juice or any juice with yeast and you measure the production of CO2 and the production of ethanol over time flies like the ferment when CO2 production is maximal and this kind of makes sense so if you look at the common features chemical products of fermentation CO2 and ethanol really stand out because of the very high vapor pressure and they're very very light so for a cue that's going to carry along distance there's really what you'd expect you'd want to choose but of course you have to pay attention to the literature and now with all these open source like fantastic things like bio archive you can get information very quickly of course my favorite one is Reddit and we came across this paper in Reddit a few years ago our house has a fruit fly problem found this in the airlock today so this is somebody brewing so these are all dead fruit flies and then the thread is kind of human yep they're attracted to CO2 because it's released by riding fermenting fruit unfortunately this makes them a really common vector for infection since they carry acid or bacteria I would postpone any further brewing until you get the shed under control okay right so the fly guy here building on that you golden death said they actually show avoidance to CO2 because it's commonly used in a version learning studies here's a recent paper on avoidance behavior the attraction so this is just highlighting a fact that within the literature there is a series of papers mostly a very high profile journal starting with this paper from David Anderson Seymour Benzer I'm sorry and Richard Axel showing that flies are aversive to CO2 okay and Leslie has done beautiful stuff isolating the receptor that's responsible for that aversion and I can't go into all the papers but they're all solid papers maybe with an interpretation that I'll question there is some complexity so Sarah Wasserman and Mark Fry's lab did some experiments with tethered flies and claimed that walking flies avoid CO2 but tethered flying flies seem to be attracted to CO2 but they got a lot of pushback on this paper so it's very very confusing and also just remember we just for kicks and giggles decided well you know how much like what are flies living in the idea in the original Anderson experiment Tristan mentioned this idea of a Shrek stuff that's something that a stressed animal gives off that other animals can respond to and that was the idea that CO2 is the Shrek stuff so if it is flies are freaking out because they are living a typical fly bottle they're living at like 1% CO2 which is kind of interesting to think about it if it is a Shrek stuff but the other problem is that if you look at insects broadly you know pretty much any insects that's been tested for CO2 is attracted to CO2 and it's what life smells like right in a sense okay so we've been studying this problem for a while and so initially we did this in a wind tunnel and we have a nice calibrated wind tunnel where we can calibrate odor plume and these initial experiments I'm going to show you work with a laminar odor plume where we can use a robot attached to a PID that maps the plume in space and then we put the flies in the flight arena and I'm not going to talk about those but flies do the classic surge and cast behavior where they sort of follow the envelope of the plume over time for minutes or hours at one point I did want to make because I just think it's kind of interesting in the context of turbulence and so forth that one of the claims we try to make in that paper is that really the time history of the flies experience is not as determined by the odor structure of the plume but by its flight dynamics because the fly keeps zipping in and out of the plume and that's what's really driving the fact that it doesn't get any odor for a long long time and it's not because the packets in the plume are moving but because it's just way out here so the locomotor dynamics are important but to take a message is flies do the classic cast and surge so one final thing I want to say though is that we found that when a fly comes across an attractive odor in the plume it becomes attracted to visual objects even when it comes out of the plume for about 10 minutes after getting an attractive odor it will be attracted to little black projected circles on the arena and I can kind of show you that a little more convincingly in this animation this is animation of data it's not a simulation and these are not all traces that were done at the same time I'm just overlaying them so you can get an impression so these are flies that are flying around there's no odor down here at the bottom of the tunnel but you can see that these flies all get attracted to this visual object and there's no odor there it's just a black projected object no no no I'm kind of compositing all that so you can see it very easily we have our Eiffel tunnel but now these are different experiments to get the CO2 problem so what we have is a little platform that the CO2 or other odors uses from and we track the flies and then they can land on the platform and then we can later track them as I'll show you so here's just I think roughly 40,000 trajectories in each experiment but this is just a heat map the flies are really attracted to this platform and they show this high activity around this dark spot and so we see exactly the same thing with CO2 so the CO2 the flies are very very attracted to the CO2 they're not attracted to just a H2O plume and we can quantify it by measuring where each trajectory goes the approaches to the pad landing on the pad approaching the dark spot both CO2 and ethanol are attractive to the flies so Mark Fry proposed in this paper these animatic experiments Sarah's paper the idea was well maybe we know that flying during flight there's high octopamine levels in the fly but not when they're walking and so maybe that could explain things so what we wanted to do was an experiment where we could kind of track the fly coming in and then test it while it was on the ground and so we built a slightly different pad that has a machine vision system so we could track what the fly did once it landed but as you can see that the flies are still attracted to the CO2 as to the ethanol after they land and that can be quantified a bunch of different ways in terms of the time on platform distance traveled approaches to odor source and so forth now I know what you're all thinking this is all a concentration effect it's all a matter of what concentration use this is not a matter of what concentration use you can use any concentration including 100% CO2 the flies are attracted to the CO2 they land on the pad and then they die so these are just two little dead flies that killed themselves going towards the fly so at this point we needed a way of doing this much more systematically so we go back to a pure walking chamber but one where we can collect data over a long period of time just to convince skeptics so here's a chamber that has a perforated floor and a perforated top and when this will turn to red CO2 will sort of ooze out very gently out from the floor of the chamber some of it will spread a little bit in the chamber most will just come out through the ceiling but you'll see what the flies do when the CO2 comes on so I don't know what else to call that but attraction so but when we set these up we're able to have the same flies in the same chamber for 24 hours and you see these patterns of activity and so this is important because this is the slide where the people who published before they're right in a certain context because afternoon when flies are not active and night when flies are not active so these are data looking at the number of flies at the CO2 or the test odor compared to the number of flies over the control odor during the afternoon at dusk night and dawn and at dusk when flies are actively foraging they're attracted to CO2 in the afternoon when they're sort of in siesta mode they avoid they slightly avoid CO2 and sort of similar results at dawn and night so maybe this is all explained by circadian rhythm so if we and this is all the clean air controls but if you starve the flies before putting them in basically that makes them more active because they've got to find food they don't go into as much of a siesta at night and you see that they're attracted to CO2 at night so it's not circadian time per se it seems to be activity and this I think is the real secret to maybe why the literature is a little bit confusing because CO2 in many experiments is often delivered at a high flow rate but as shown by many labs including David Average's lab insects tend to not to slow down and become inactive when subjected to high flow because they could get blown away it makes most sense however if we take that so this we modified our chamber to really kind of blast not really blast but we increased the flow rate by a factor of about 5 and now the CO2 becomes aversive however if we cut the arista off so they can't detect the high flow rate they're attracted again or if we do the high flow rate but we heat the crap out of them so that they're really active CO2 is attracted again and so this was sort of the secret this is probably the most important for the sort of CO2 story this is just a whole bunch of experiments that's parsed according to the mean locomotor rate before we present the odor from low centimeters per second velocity to high velocity and then as you go along you can in each one of those experiments you can calculate a sort of preference index where they attracted to the CO2 or aversive to the CO2 and color code that and you can kind of do the statistics and you get a nice linear regression of this preference index as a function of velocity or you can just sort of parse the data below a certain velocity you can look at their responses remember this is the pre-stimulus velocity above a certain velocity they're attracted so in the same basically population we can either see attraction or aversion depending upon the activity rate of the animal and then this is really for aficionados like Leslie because now that we kind of know how to see attraction and aversion in the same animals in a cohesive way we can start to go back and see to try to chase down the receptor and their important results is that the receptors that people thought were responsible for aversion are but they're not responsible for attraction and you know these are just very intermediate preliminary but so far we've sort of chased it to one of the IRs that's a co-receptor so it functions start to function with another receptor a protein but if we get rid of this the flies are aversive but they don't show attraction so it doesn't look like the same receptor is used for the two behaviors use a different reception so the other I just want to say two things about this one I mean it's bad to say ill of somebody who left the conference early but Bill made this comment that maybe everything in the fly brain is parsed into like attractive aversive attractive aversive and that's all mapped in the lateral horn I think I think flies are smarter than that I think they have the ability to see the same odor as attractive aversive under different context and I think there's other evidence in the literature that supports that another thing and this relates to a giant ecological model that I'm not going to have time to talk about but what this is plotting is the number of flies on the platform you know that landed on the platform that had CO2 after a while they start to go away and that extinction curve is indicated here they do the same thing for ethanol but they extinct at a slower rate these are the same data collected in this internal chamber that is sort of once the CO2 turns on they all get excited about it but then they slowly it doesn't become aversive but it's just not interesting but this totally makes sense like you smell something and you look in and after a while it's like I'm screw this there's nothing here right and that decision is made earlier for CO2 than it is for ethanol but that makes complete sense because the environment is full of lots of false signals of CO2 much fewer false signals of ethanol there's really nothing but yeast out there making ethanol and if I forgot to say this flies are yeast divorce right that's the whole deal okay so why should they avoid CO2 I you know with all due respect I don't believe the Shrek stuff hypothesis the reason very simple flies are tracheates that store CO2 in internal tubes in their body until it accumulates and when they breathe they give out big pulses of CO2 that's how they breathe and this is why I think if you vortex flies you get CO2 so I don't see any way that like CO2 could be used as a stress signal because every time a fly took a breath it would be interpreted as a fly freaking out right so I just don't think it would work okay then why do they avoid CO2 there's two hypothesis we come up with why is the volcano hypothesis now as crazy as it is there's a literature on this that entomologists have found cases in geological areas where there's massive insect die-offs because of an increase this is also killed cross-country skiers in places that CO2 activity increases and all these insects get attracted to it and then they die so there could be some aversion for natural sources of CO2 I think this is a little crazy but it's just should be mentioned because there's a literature on it I think a much more likely hypothesis the parasite hypothesis that if you're trying to find if CO2 is what life smells like and your fly and you go after the CO2 and you're something that likes to lay your eggs in the fly or it's eggs or it's pupae or it's larvae why don't you go after CO2 as well and when I said that somebody asked me how do you know they're your flies on the traps well we never collect fruit flies in Coyote Lake but we collect a crapload of stuff and almost all of them are parasites so you saw in the middle of that wasteland these are just sort of a small number of these little tiny insects that we collect so I think that basically you know when flies are quiescent and they're not actively foraging they don't need to be near CO2 they don't want to be near CO2 because you know that the parasites could come get you so that's our kind of best hypothesis so how much time do I have left like five minutes okay now I wish I hadn't cut those 20 sides no so I want to talk about another thing that is like so I mean I'm kind of cheating because this doesn't relate to odor except in a way that it does but this is in the context of narrow search so now we're in a very very different spatial scale so you know the flies possibly flown kilometers zipped around searching and I didn't have a time to talk about that aspect but it makes a decision to land and now it's looking for food what does it do well you know a guy who I've admired a real long time is Vincent Deteer published a book called The Hungry Fly and a charming book called To Noah Fly and one of the phenomenon that he described and he was just basically a fly whisperer who was really a pioneer and sort of invertebrate neurothology back in the 50s but he would take house flies and pick them up and give them some sugar and then put them back down and then draw what they did and they did these kind of crazy loopy things that he called dances and he called them dances because he reminded him of the crazy loopy things that honey bees did when they come back to the hive and he didn't use the term lightly he actually proposed that it was the same like ancient behavioral module of what an insect does after it gets rewarded by food now I actually think that specifically that's nutty but generally it's brilliant because I think he was a very very early pioneer in this idea that you could observe behaviors in radically disparate species that actually went back to some core ancient the Devonian toolkit so I have enormous admiration Bell looked at this in fruit flies a couple of decades ago but again not with technology that I think made it possible to figure out mechanistically what's going on so Irene Kim took on this project in the laboratory and I just want to show you some movies of this so here's a hungry fly walking around in a dinner plate this is sped up a little bit this is the timer there's a yummy yeast paste drop in the middle of the arena so this fly is hungry point number one it takes an excruciatingly long time for the fly to find the damn food it's like standing behind the person in line and then they lose their wallet and you're like you know it's just this is moments after the fly has finally found okay this is that same fly that's what it does okay so this is this thing that Vincent DeTier called the dance and the referees of the paper didn't allow us to call it the dance which I think was really unfortunate what's that no no no I wasn't so this is just looking at lots of data you can do this in complete darkness and they still do the dance you can buy a number of different ways test whether odor is involved and using you know sucrose using mutants although the mutants are somewhat problematic for reasons I can get into for aficionados but I'll show you a more convincing experiment a moment and an interesting thing one idea is maybe the flies are laying down a trail so you can get rid of the hydrocarbon glands these enocytes that females have that would be the most obvious things and would sort of lay down some sort of track and in fact the dances if anything get super strong if you get rid of their ability to lay pheromones in fact it's so amazing I just want to show you the results of that experiment so again this is a fly that has a genetically ablated enocytes which are epithelial cells that can secrete hydrocarbons that are used in signaling so again you know the excruciating weight for the fly and if alfaction is so important why doesn't the fly find the food okay so this is the dance right after it's just crazy to me okay so I hope some of you are still not convinced I mean what we're trying to make the argument for is that this is entirely idiothetic that the fly is doing this because of you know keeping track of internal odor commands or sensory you know mechanosensory information so of course what you want to do is the disappearing food trick or the sliding food trick so here we create an arena where the food is on a little strip and the fly can find it and it starts dance and then we sort of you know we move the food to a different part of the chamber it's a difficult experiment to do because flies have such great mechanosensory capabilities that they tend to get disturbed by the sliding but nevertheless it still works so here and there's something just ridiculously amusing about these and these are done in the dark as most of our experiments are so again you know there's a long wait while you know the person like looks for their their wallet and forgets they need bananas okay it finds the food and then you'll see the dance okay and then you kind of have to wait until the fly and then see it was moved over there and the fly continues to to do the dance and I should there's lots of features I'm kind of skipping out that the dance the loops of the dance get further out over time and so forth so this okay this is just okay this last trajectory but it's also kind of cool this is a case where the fly finds the food in the new location and and so it's got to find the food come on come on okay there finds the food and then it'll do the dance and then we'll move the slider and then this is just a fly you know about 30% of the flies do find the food in the new location and then you'll see like what it does immediately afterwards the food's moved the dance is around the old spot oh finds the new food and now it starts to dance around the new food so you know whatever the food is done it's just like reset this homing vector and so these are some of the yeah so this is just a series of experiments and these papers out I mean but you know the point is this is you know the searches the behavior is still centered even after you've moved the food so it can't be that they're going back to it because of the odor because it's you know it's somewhere else now and then just for you know the mathematicians you know this is the analysis like could it just be monkeys and typewriters so we take all the statistics of the path lengths the turns the turn angles and you know you can create any model based on the locomotor statistics and you just can never recreate dances that are centered around the food so there's no easy random walk model that could explain the behavior another analysis that we've done is if we plot the distance to the food over time you see this sort of periodic thing because it's going back to the food and so if you do a auto covariance analysis of that you see these sort of like side bands that indicate there's a periodicity that you see in the like random data and that's kind of interesting because you know it would suggest that the flies have some sort of timer and okay so those who are insect aficionados will know that Categlyphus you know this desert ant has this amazing path integration ability where it can leave the nest go out looking for food find the food and go exactly back and this is like the first grade evidence classic evidence that flies ants have this idiothetic path integrator you know the classic work of Ruder Gravener and then Matthias Wittlinger did these amazing experiments where he glued ants glued stilts to ants so that after they found the food they thought they traveled further than they did I mean or you can see so you screw up the idiothetic sense and they overshoot or undershoot depending upon whether you you glue them to stilts or you make little stubbies by cutting half their legs off so we think that you know this is part of the Devonian toolkit that Drosophila can do this as well so the fly finds food and now it's sort of keeping track of its orientation and heading and then at a certain point it says okay now I've got to go back now I'll go out again come back go out again and so forth so I think the interesting thing to me that relates to odor can bring this back to odor is that it really was excruciating that the fly had such a hard time finding you know it's like favorite food it's like caviar in the middle of this dish but I think it actually makes sense because I think that there's no there's no plume because there's no we're not we don't have advection within that chamber so it has to rely purely on diffusion and I think flies are just really crappy about doing that I think and then I think it's just really actually it's kind of harder to do chemotaxis once you're here than it is once you're in the air which is why I think that the fly like once it finds the food there's been really strong selection to like remember where it is because you know so I'm not saying that like animals don't need noses I mean my whole point is that what's really cool there's something cooler than like a fly of a flying animal but you put a nose on that flying animal and then you have just an amazing thing you know there are limits depending upon you know upon the physics and so forth and I guess the other thing I have to say maybe there's more for discussion I think what one thing that this I hope this work points out is that all these behaviors although they rely on olfaction they also involve you know mechanoreception and vision and you know as sort of a sensory motor person you know I think sometimes it's a little dangerous to sort of like only focus on one sensory modality because you know from animals behavior from animals perspective like you know it's squeezing all the sensory information from the world you know like blood from the turn up and sometimes you know they're interesting interactions between olfaction and other and other senses but anyway that's basically it and again I want to highlight the people Kate who in a very short period of time has like gotten these coyote lake experiments to do which is just really amazing Flores Van Broegel has been chasing down the CO2 story for several years now and I think you can get a sense of just how hard it was to figure out what was going on and Irene Kim did the work on the on the dances so thanks