 I'm trying to organize snorkeling in the marine reserve in front of the castle the world wildlife foundation runs Provides gear and things like that So I was thinking trying to do that Friday afternoon We need at least six people for them to come out with the gear and all those things And it's 20 euro per person And I think they show up with a guide and all sorts of things So if anyone else is interested I think and I don't know after this is over So two o'clock or something like that Just send me an email or grab me a coffee and and I'll get it all set up. Thanks Okay, so I'd like to introduce our first speaker Whose name I can't pronounce he keeps correcting me and I keep getting it wrong. So here goes Kaylin Vassigian who's gonna tell us about evolutionary dynamics of communities of antibiotic producing bacteria Thanks Peter. Well, it's it was better not perfect, but It's Do you hear me well is this thing on Okay, so first of all, I'd like to thank Daniel and Simon for Inviting me and also to all the other organizers Mattel Vittorio and Pan June for putting this incredible experience together Now I'd like to honor the place we're in which is the Center for theoretical physics in my talk Just like in the last talk Yesterday, I will emphasize more theory and I also sprinkled it a little bit with physics Way of thinking along the way Okay, so let's see if this is working. It used to work. I don't know. I'll just maybe use this Okay, it's working. I was working. Yeah Okay, so the story here, I'm going to tell you about started by thinking about Extremely diverse environments such as soil where people say that you can have 10,000 some people say 40,000 species in a single teaspoon of soil Okay, and I want to think about how diversity is generated and maintained in environments like that okay now Now there are at least two big Potential classes of explanations for diversity the first type of explanation is that the environment such as soil Are heterogeneous down to small spatial scales? Okay, and therefore you have many different abiotic niches there and correspondingly They harbor many many different species, okay So basically the diversity is driven by environmental heterogeneity the second type of explanation Which is not mutually exclusive is that the diversity is somehow maintained through a network of pairwise Or not pairwise the network is some kind of web of interactions. Okay Yes, we're going to talk about about that We're going to consider kind of the heterogeneity created by the microbes because I'm kind of Yes, I said at least two explanations there are other types of categories, but one obvious one is that Kind of there's some environmental gradient, right? It's not like a big puzzle why the power bears and the grizzly bears coexist because they there's they have it different environments, but Yeah, this is not a full list of explanations right the microbes Organisms create niches for their organisms. Let me start. Oh now it's working perfect. Okay So doing theory you cannot weigh the relative importance of these potential explanations But what one can do is kind of ask is this second type of explanation Where the interactions maintain diversity is even plausible Does it make sense is it self-consistent? And if it does kind of offer some particular mechanisms by which this can happen beyond a simple hand waving Okay, now we're going to focus on this second Category in order to eliminate the first types of explanations we're going to ask can we get emergent diversity in one niche environments and for the purposes of this talk What I mean by one niche environment is Let's say single food source no abiotic spatial heterogeneity though we can might have an emergent biotic heterogeneity, okay, and basically we're going to ask whether the situation where we have one species per one niche is Evolutionary unstable with respect to emergence of some diversity. Okay Now, why would that be? Well, one reason this might happen as it was discussed yesterday actually is that microbes do not simply Occupy niches created by Someone by God But they actually actively remodel their environment. Okay, and in this way they create niches for their microbes and kind of the primary way in which microbes do that is by secreting or removing bioactive molecules From the environment. Okay. Now all microbes do that because of the very minimum they need to eat and they need to secrete waste products and As it was mentioned yesterday Waste products secretion already gives us one mechanism for coexistence, which is cross-feeding. Okay It is a very well known mechanism, but perhaps what is slightly Less appreciated is the extent to which you can get diversity to cross-feeding and I think this is the subject of the next talk But for the purpose of this talk, I'm going to eliminate This known explanation. I'm going to shoot myself in the leg and said Let's imagine that waste products are just waste products and everyone is just eating the same food Can we still have a diversity that says coexisting here Stably in that environment. Okay, and the reason I'm interested in that question is because many microbes invest heavily into the production secretion of Bioactive small molecules such as antibiotics such as a their force quantum sensing molecules and so on Now we know quite a lot About the chemical structure of these molecules how they're produced how they act But we're not very good at understanding their Ecological and evolutionary consequences. Okay, and it's hard for several simple reasons first of all The effect of small molecules is kind of context dependent simple example antibiotic production might be Beneficial if there are sensitive bacteria around but it will be just a waste of resources and energy if there are only resistant bacteria around and perhaps because This is a benefit is context dependent many microbes have Evolved the capability to produce many different Antibiotic like molecules. Okay in some cases up to 70 different molecules. Okay, so what you see here It's it's basically for you know bacteria. How many? Biosynthetic gene clusters they have for things like a their force and antibiotics You see sometimes it goes up to up to 70 typically around 20. Okay, and And Kind of so that's an extra complication and the final ingredient that makes the dynamics very interesting for me Is that these small molecules mediate ecological interactions and these interactions can evolve really fast through several different mechanisms? I just told you that microbes can produce Many different small molecules. They call that the capability, but they don't they don't make them all at the same time so simple regulatory changes for example can Change the subset or quantities of molecules you produce right which of which antibiotics you make also the response of Bacteria to small molecules can easily evolve with antibiotic resistance being a famous example of that And finally both production and resistance are very modular Genetic traits so they can move Very easily through horizontal gene transfer. Okay So to summarize what I'm telling you so far. We have systems where we have strong Ecological interactions this ecological interactions can evolve fast. Okay, and This goes to the heart of the problem. I think we have with understanding Microbial communities in general this coupling between ecology and evolution. How do we think about it? Okay, I kind of when I think about ecology even though it's very complicated I have this picture of a non-linear dynamical system with some fixed dimensionality and you have some expectation that These trajectories can either reach a fixed point or maybe exhibit oscillations or maybe Deterministic chaos or something like that, but when you add evolution then it becomes very confusing because what do you do every time a New strain emerges do you add a new dimension every time a strain gets extinct? It's it's very unclear how to think about equal evolution. Okay, so this talk has kind of two teams one is Diversity which is how I got started and then the second team is equal version of the dynamics, which is kind of The way things evolved and kind of what's most exciting for me right now All right, so within this setup We're focusing exclusively on antibiotic interactions and we can ask questions on two levels. The first level is Entirely the ecological level and this is a work that I did with a Roy Kishoni and the student graduate student Eric Kelsey and In that part we focus on the question can we construct communities kind of in silico Where diversity is maintained? Where diversity between Organisms that have different anti-pollution production and resistance capabilities maintained in this one niche environments. Okay, and then we'll add evolution to the picture and ask If you can construct such communities does diversity emerged spontaneously if you allow microbes Kind of in a natural way to evolve their production and resistance capabilities okay, so First of all, it's not at all obvious that antibiotic interactions help diversity So it was not until 1997 that the mechanism was proposed for how this might even happen and in fact the person that proposes is Simon Levine and he also mentioned in in his talk and he in his talk he also kind of acknowledged contribution from Bruce who previously studied combinations of producers and sensitive communities of producers and sensitive bacteria and compared liquid versus Solid media So kind of I might summarize this by saying that Simon Levine Stunning on the shoulders of giant Bruce Levine kind of Recognized that Antibiotics kind of naturally very naturally lead to rock paper scissor games. Okay in the way this works is The producer kills the sensitive bacteria the sensitive bacteria out to be the resistant because of cost of resistance And then the resistant bacteria outcompeted the producer because of cost of production. So this leads to Diversity by the way this works is it only works in spatial setting and basically you have these different Single species patches that start chasing around like the scissors chase the papers and the paper chase the rocks And you have these beautiful spiral ways waves but as Simon And other people later realized This only works if spatial structure is very carefully well preserved as soon as you start introducing kind of microbial diffusion or microbial dispersal Diversity collapses, okay So this kind of limits the utility or it apparently limits the utility of this mechanism for explaining natural diversity In soil for example If dispersal is high, yes And this oil for example this antibody producing bacteria that we work in the lab There's a sport forming their hydrophobic spores every time it rains these parts will be carried at large distances So even in spatial structures preserved temporary in the long run over the course of tens of years It will be a bit scrambled So this leaves us with this more Pointed question, which is can we have diversity? Without perfectly preserving the spatial structure at all times, okay and kind of the way We made progress with that question the inside came from an experiment I did with Eric where we looked at higher-order interactions in bacteria And what we discovered is that a month bacteria from the genus Sceptomyces at least antibiotic degradation is a very common higher-order interaction. So basically many bacteria Degrade the degrade antibiotics and and this is ecologically interesting because This antibiotic degradation medias higher order interaction the degrader cross protects the sensitive Bacteria from the antibiotics or sensitive bacteria can survive near an antibiotic producer if there is also an antibiotic degrading bacteria near it Okay so We put this in a very simple mathematical model and Basically, we have different Strengths with different phenotypes. We put them on a plate the producer secret antibiotics in a certain neighborhood Degraders destroy antibiotics in a certain neighborhood if there are sensitive bacteria covered by antibiotics They get killed and finally We kind of collect all the spores from this virtual plate mix them up They loot and plate again. Okay, and so we work in this perfect mixing limit And the reason we do that is that's because this is the most interesting limit from a theoretical perspective given this limitation that I Explained to you. Okay, and also thinking ahead about experiments. It's a very natural way to do experiments You plate spores on a plate then you collect them Okay, so to make a very long story here Short what we discovered is that if you add this extra ingredient antibiotic degradation rather than just a regular resistance We can in fact construct stable coexisting Communities where the strains only differ in their antibiotic production and resistance capability and eat the same food whereas If you have just pairwise inhibitor interactions without cross-protection we can never Construct such communities no matter how we play with interaction network. No matter how we adjust the growth rates and when I presented this kind of I Think it was a year and a half ago Jeff Gore made this comment that well, it's There definitely it is definitely possible to have stable coexistence with just pairwise rock-paper-scissor games and While this is in general true kind of the constraint we have here is that we're ignoring The possibility of predator-prey interactions basically we assume that the producer and the sensitive bacteria They eat the same resource they compete for the same resource So the producer bacteria is not better off if there are sensitive bacteria around So no single strain is bringing extra resources to the table, okay So within this constraint kind of this Statement is is true. Okay. It's impossible with this pairwise interaction So only to have this table communities, okay And this here shown just one example of a kind of nicely symmetric community with inhibitions and Degradation interaction show once dashed lines as you can see Here what this triangle shows is that if you mix the species in many different abundances The dynamics nicely close to a stable fixed point here. So there is a big basin of attraction That particular community and kind of the other nice thing is that if you start playing with the growth rates The relative growth rates of the species this mechanism is robust to big changes in growth rates We can have coexistence even for like ten fault differences in the growth rate between microbes Which is kind of huge for microbes like one micro growing ten times faster than the other All right, so one thing that will be relevant in a little bit is that kind of the simplest Community that we could construct within this framework Maintains coexistence between three species on two antibiotics and all the resistance is coming from antibiotic degradation, okay, and kind of just to Connect to with what Jeff Gore was saying Yesterday about these three different types of pairwise interactions dominance and by stability coexistence it turns out that even though We kind of started with this picture of rock paper scissor games these communities that we find in fact Not like classical rock paper scissor games where you have three dominance interactions in a line But they are what I would call conditional rock paper scissor game. We have Dominance here dominance here, but this is kind of by stability. In fact one can prove mathematically that you have You need to have at least one by stability for these communities to work. Okay all right, so We're super excited about this result but There were few outstanding Issues that I wanted to address before kind of fully embracing this mechanism as a viable option for explaining diversity in Natural environments kind of the first limitation is that as you try to construct bigger and bigger communities It gets increasingly harder to choose parameters that Have coexistence So it's not at all obvious that evolution Continuously will find these strange Communities that coexist. Okay, so that's kind of the big question if you let if we let evolution naturally happen Whether it will find these communities and finally kind of this model that I showed is very stylized and that by design You wanted to want it to be very simple, but there is always a danger that if you simplify too much Maybe what you see is an artifact of some simplification. So we wanted to add realism stuff that we know Is happening from our experiments in the lamp and see if the mechanism of Maintaining diversity through antibiotic production and degradation will still work if we add some realism so kind of one of the realistic features we added was The life cycle of first step to my sis So they start the spores these parts germinate they form mycelium colonies and As resources get exhausted They Sporilate okay, and here at the onset of sporulation. They usually flood the area with antibiotics and that's because What feeds the sporulation process is that mycelium commit suicide? And the released resources are used to to build the spores, okay So they're not as smart as fungi maybe to kind of internally incorporate these resources a the resources leak out into the environment through So it's kind of a protection mechanism. You don't want someone else. You're unable to steal your resources So there are three essential features. There is growth There is antibiotic production and then there is a sporulation phase Okay, and here is how it might work in a simulation you scatter some spores on your virtual petri dish the colonies grow and they crash into each other and By the way here in this way of doing things basically the biomass you get per colonies exactly What will was saying it corresponds to exactly to the what is it called the? Voronoi Desolation and then different Colonies have different properties. So what's shown here in the red red contour serve antibiotic producer colonies So they produce antibiotic that starts diffusing what you show them here in say on our antibiotic degrading colonies They start degrading colonies and kind of Over time you start getting this many different micro environments Okay, just because of the randomness of who? Who is neighbor to whom and then at some point you have a sporulation phase Basically what? We assume that if there are too many antibiotics you die and then There is higher a sporulation If your neighbor is dead, okay Basically your neighbor cannot steal yours your resources Maybe if you kill that neighbor, maybe you steal some of his resources and that's actually an actual feature You see on petri dishes if you have a zonal inhibition usually around the edge of the zonal inhibition You see better growth and better sporulation. Okay, so this is where the benefit of inhibition comes This is where how it inhibition pays off in the model and That's kind of important because for evolution you need an immediate benefit of inhibition. Okay? Yeah, so each colony Secrets antibiotics this antibiotic starts to diffuse away just just regular diffusion then Degrading colonies they act as a sinker to the antibiotic. They just destroyed antibiotics. I don't assume that The antibiotic destruction think diffuses away just like the colonies Start destroying antibiotics then you assume that there is like a standard zone of standard response curve So if this is the antibiotic Concentration and this is kind of log survival. So there is something like the analog of MIC if you want in our model and after that there is an exponential decrease in the probability to survive per unit time and that's that's it and then after Some time we just said okay, and now we're done with that now we're going to do sporulation And you'll sporulate proportional to How much life cells you have? But there are also these But then you assume that there are these background of resources created From dying mycelium or from killed bacteria and these resources started diffusing Okay No, it's by antibiotics That are produced by the producer colonies Does that make sense? Okay, we can talk a bit more later Yeah, the antibiotics usually are degraded by enzymes and there are two options. Either the enzyme is diffusing Or it is linked to the surface. I'm kind of assuming it's linked to the surface So it was degrading the enzymes is not really diffusing. It's kind of the more conservative assumption if you Know that degrading enzymes stay stuck to the cell wall of the Of the bacteria that degraded at least in this model. I mean it's you can easily change that with one line of code Okay, and kind of most importantly you can add evolution and to do evolution properly We kind of go beyond these discrete phenotypes degraded sensitive resistant producer and kind of assume Continuous possibilities of different phenotypes in particular You assume that you can adjust your level of antibiotic degradation continuously You can adjust your level of antibiotic production continuously and we know the gene regulation can do These things so it's a very natural assumption And kind of on this branch here We assume that on the blue branch bacteria also equipped an efflux pump Which is kind of to simplify things and don't And not introduce an extra dimension. We assume that the efflux pump Kind of is feedback regulated and automatically brings the concentration up to this point To avoid any inhibition. Okay Yes Yes, I deliberately Assume that just because that allows me to have a one dimension per antibiotic Otherwise, I have more dimensions and it become more complicated But we assume that you cannot for example produce and degrade antibiotics You cannot have an efflux pump a db degraded that simply to reduce the complexity of the problem But you agree now that we have some interesting stuff in the next thing I can relax this and see all this all this good thing still working if I realize that assumption. I haven't done that Okay, and so we have this in these simulations We work with either one or two antibiotics and then you have a Each point in this space Is kind of a possible strain And then we assume that you can have mutations that you can either tweak The phenotype with respect to one antibiotic or kind of drastically change the phenotype And I showed you in the beginning some mechanisms to which that might happen For example to jump from here to here. Maybe you can turn on a silent antibiotic cluster or you can acquire antibiotic from someone else And since such a low dimensionality actually doesn't matter what the mechanism actually is what we assume Is it's easy to make small changes and it's easy to make big changes of antibiotic production and degradation And in this simulation we started a single strain that sensitive to all the antibiotics Okay, and then we just run Thousands of different simulations with different cost of production and resistance And so on and see what happens So let's start with the one antibiotic case. So we have a single Uh phenotypic dimension here. So this is the space of possible phenotypes And we also kind of color coded them all the producers are magenta all the degraders are green the sensitive limit is black We have resistance in blue And we run for let's say 30 000 ecological cycles and I kind of zoomed into the first 500 cycles because Uh things happen fast And what you see here what I wanted to see is that kind of very quickly in a matter of Let's say 100 few hundred ecological cycles you get these three species communities forming So it's a three species communities between a producer strain Uh sensitive strain and a degraded strain So it's uh like Community the simon proposes but instead of a resistance strain you have a degraded strain But now it's spontaneously emerging Okay, and moreover it's evolution. It's evolution is stable state and what I mean by that I mean it's ecologically stable if I turn off mutations it will persist indefinitely and you can see the nature of the tractor here And second of all no mutations exist that can destroy the community Actually in practice what happens is even after a while you can have fine tuning mutations here One producer dies off another comes in So the community keeps fine tuning itself very very slowly after a time But uh, uh, it's stable. So this is like a Nash equilibrium between three species that emerges and persist Okay, and it kind of snaps out of nothing Now, uh First of all see that Kind of in what I showed you before with the simpler model. We had we had We we required two antibiotics minimally to maintain three strains Here we have coexistence with only one antibiotic of three strains which is emerging spontaneously But kind of if we if we focus on just on the pair wise interactions again We see the same pattern. We have a dominance interactions dominance interaction Then we hear this by stable interaction and this by stability just emerges. I mean, it's not for any Values of production degradation. You have by stability just that wherever we see This evolution is stable state emerging you also always see this this by stability And you can understand it because both antibiotic production degradation They're like cooperative mechanisms in this model. Okay, you might be a lone producer And that might be insufficient to inhibit a degrader But maybe if you find there are five producers surrounding a degrader, they can kill it Similarly with degradation like a lone degrader degrader might not be strong enough to protect itself But if there are many other degraders that might help them, okay So, uh, again, uh, we don't have a classical rock paper scissor game. We have for this kind of conditional Yeah, rye Yeah, there is a Is there a cost of production? Yeah, we assume that the the more you degrade There is a Linear cost of degradation. There is a linear cost of production. We also imagine there is a cost of having this efflux Pump mechanism another cost to we have is an operational cost for the efflux So resistance in this model if you have an influx plant is not just like a constant genetically in quality It depends on the antibiotic concentration of the environment kind of basically what you assume is that So if this is a cell and there's some efflux pump, so this is the outside antibiotic concentration This is the inside concentration So we imagine that you expand some Energy which is proportional to some constant Time to and this by the way the inside concentration we assume goes to whatever is Non-inhibitory So basically you have this operational cost for the efflux pump And we all these costs all these different cost parameters to the model in which Yes, yes, that's a Of course true. Yeah, if you have a compensatory mutation, for example, you cannot get stability I mean every if for example the ability the possibility of compensatory mutations exist Obviously every antibiotic will get absolute over time and you never get any evolution stable community, right? So it's very important and actually I think Martin also has In his talk later he'll discuss situations where maybe there might not even be a cost to production and stuff like that We assume there is cost to resistance. We assume there is cost of production here. All right so kind of To connect a little bit again with what jeff was telling yesterday about no rock paper scissor games existing And so I decided to show you a bit of our data So we did many experiments similar to jeff's where we mix many step-to-mises in the lab and propagate them just kind of on petri dishes and monitor the outcome and unlike His collection of strains this random soil antibiotic producing bacteria from the genus leptomises They very often exhibit by stability. You can see it for example here. You can see it here It's kind of very striking you put one strain at 60 percent the other 48 wins then you flip it the other strain wins But then just by trying few random triangles three combination strains you already can see that kind of this dominance dominance by stability this weak Or conditional rock paper scissor game already emerging in terms of the pairwise interactions And just like jeff we never see this dominance dominance dominance rock papers is wrong game So obviously here in this community that we see the higher order interactions are not correct to Sequel existence, but at least looking at the pairwise interactions. There is nothing that discourages from thinking that there is something wrong about our way of thinking all right Now we can do two antibiotics and Here is what you see now with two antibiotics You have a two-dimensional space We plated we look at it over time these are the strains above 0.1 And very very quickly actually you cannot even see it on this time scale of five species Evolution is stable state emerges. Okay, you can see how it emerges again in a matter of a few hundred generations and then it stays on And we have covered all the different phenotypes to help us look at this and you see some selective sweeps keep going this community Refines further and further but this is an evolution is stable state once it forms it stays there forever There are no mutants that can exist. So it's an example of nasi equilibrium Evolutionary nasi equilibrium that emerges very very quickly and kind of snaps so again The this is the kind of complexity which is much higher than what we've been seeing In this much simpler model when we were trying to manually construct communities here we get much greater complexity for free All right and With the single antibiotic there was a single evolution stable state of the type I showed you here We have two different types of evolution stable states for different parameters. So there's no by stability. It's just like for one set of parameters we get This the type of community for others we get this type of community and just to orient you a bit Here we have a antibiotic producer to one antibiotic Production of another antibiotic degradation of one antibiotic regression the other antibiotic and this is kind of a double sensitive strain For other parameters, we have the same thing but instead of a double sensitive strain. We have a double degrader Here and there is some kind of modularity to this community If I take these three strains with one antibiotic They form again this producer sensitive degrader. This is ecologically stable It's not evolution is stable obviously, but it's ecologically stable. This is also ecologically stable So we have this kind of one antibiotic modules that stuck together somehow To make these bigger communities and you can keep going like with three antibiotics. We we can see seven seven species coexisting Okay, so for the rest I want to simplify kind of our Our notation so instead of having these two dimension three dimensional diagrams of phenotypes over time I'll project the phenotypes into one dimension like that and then we're going to look at Phenotypes in some abstract space versus ecological cycle and you can look at what happens with the different lineages Okay So first of all to summarize what I've shown you so far. I've shown you That we have emergent diversity in one niche environments If we allow the bacteria to freely evolve their investment in antibiotic production and antibiotic degradation in kind of realistic multi-scale model and moreover the types of communities we see Kind of more complex than the ones we were seeing initially so adding Realism and adding evolution not only didn't hurt our mechanism. In fact, it made it work better Okay, so this This is kind of what we were hoping to see that's why I started the project looking at diversity But then when I started looking at the data for many different parameters many different cost of production resistance in looking at these kind of diagrams Something struck me and what struck me is that as we change the parameters There are there are several Different types of patterns that you start to see I call them different modes of the equal evolution dynamics So what I show you so far are the evolutionary stable states Which we kind of know and love and know how to think about them But there are several other Equival modes and I'll go through them right now one of them Is what we call rapid turnover or red queen dynamics there you have diversity at every point in time But you don't have evolution stability ever So the so the community keeps changing and changing and changing no single strain as you see there are no straight lines here No single strength persists for very long So we call it red queen Because you need to keep evolving in order to survive No one that doesn't evolve survives for very long. Okay The next thing is kind of if you look in between the evolution stable state And it's a rapid turnover state A striking pattern we see is intermittency. Okay, so we see this Uh Kind of apparently stable communities that survive for thousands of generations Kind of of the same same type as the evolution stable communities five species community And then suddenly this community would collapse then you have random rapid turnover for a while Then you will get a kind of stable community that survives for a long time again And here is another example. We just see it over and over and over again And this is a particularly strike a beautiful example of it You see the community there are some selective sweeps happening within that community and finally it collapses then it's random random random random Finally it hits into another community then it collapses again. Okay, so we call this Intermittency Okay, so we have two qualitatively different regimes with very different statistical properties That persist for long times thousands of generations and sometimes switch from one to another Okay, and I called it intermittency because that's kind of a Famous phenomenon in physics in particular in the context of turbulence Imagine I adjust my the waterfall On my top for the low setting then I have a nice laminar flow If I increased of the on the flow rate, I'll get a turbulent behavior But then for entire range in between You get what you see in the picture kind of the dynamic stays in one regime Then switches to another regime stays in the regime for a while and switches back. Okay, so basically it's kind of a Non-phenomenon in physics that's when you transition to chaos There is this a dynamical range where you see this intermittent behavior And I think down again our case is that kind of the ordered phase is the evolutionary stable state Then we have this chaotic rapid turnover phase and we see this intermittency in between So that's one way To think about it Kind of another parallel one can make is can make is with phase coexistence in physics so So imagine you start with vapor and then you start compressing it down at constant temperature It will eventually turn into liquid, right? But there is this entire finite size regime Where both the liquid and the vapor will coexist with each other I think This is what we have here. We have a rapid one phase. We have another Kind of a equal equal phase. Yeah I'll do this guy Yeah, yeah, so so basically this This brown guys they appear occasionally in this chaotic phase Okay, but they don't start this intermittent see but once you're in the rapid Yeah, the random phase sometimes you get this this brown guys are double antibiotic producers actually Okay, so, uh, anyway, we have the rapid turnover phase. We have the evolution stable state and you have this I think the intermittency is just the phase coexistence between the two Okay, uh, we can also think about It in actual biological terms. So when you look at this long persistent communities You can see that they're ecologically stable But they are not evolution is stable. So mutations exist that can co-opse these communities single mutations can come up and co-opse these communities So why do they persist for so such a long time? Sometimes tens of thousands of generations? Yeah, very very good. Yeah So so it's it's temporal Coexistence and the reason we don't see spatial like a simultaneous coexistence is because at the end of every cycle We destroy all of spatial structure Completely, okay But I think if you have an extended Maybe you don't have the computational power right now, but if we if we had had a big spatial simulation It's conceivable And I might be wrong, but there will be one region which will be in the Kind of nice five species community phase. There will be another region over there that will be rapidly fluctuating Okay Not exactly Yeah Yes, yes Yes, yes, because I showed you there is a single evolution is stable state. There is a single type of community that can Snap together and persist for a long time. So that's why you always see the same thing Okay, so This community is an ecologically stable. They are not evolution is stable Why do they persist for such a long period? It turns out the mutations that collapse them They're always very close to existing phenotypes. So here the five species the five strains And there's some mutants that collapse, but they're very close. You have to zoom in to To see them and because they're very close to residents their selection coefficients are very tiny Okay, because the selection coefficients are tiny. They're almost neutral mutations. They're very It's very hard for them to escape stochastic loss initially Even if they escape stochastic loss It will take very long time for them to invade even if they increase in frequency They'll stabilize the community only very weakly it will take Hundreds of generations. So all these things accumulate and that's what kind of is the Proximate cause for this Intermittency but there is a very simple intuitive way of thinking about that if you're in the rapid chaotic phase They're all the time Big unexploited ecological opportunities. So the selection coefficients are huge once somehow Five species community forms it kind of Almost exhausts all the ecological opportunities there And there might be some Some beneficial mutations that might arise but But these opportunities are Not strongly selected for okay So the so this exhaustion of ecological opportunity slows down the the evolutionary dynamics All right, and kind of my favorite fixture and I have slides later to show you that is like The rapid emergence of evolutionary stable state corresponds to this funneling dynamics. So if you think of This bow as the state of the system And this is kind of the ecologically stable communities Once an ecologically stable community forms in this regime It's kind of funneled down to the evolutionary stable state Termitence is the same thing but the funnel is open down on the bottom So basically the bow goes into the funnel spends a lot of time there drops back and it keeps doing that it's kind of related to mechanisms in physics such as Long-lived excitations Let's say if you shine light on phosphorus atoms they absorb the light But then they'll go to this Other state and the deactivation will be very very slow. So we have this long Lived loops that there's like some randomness that excites the system, but then it's very slow to come back All right, so this is physics Yeah, how I just take five four minutes Okay So, uh, here's some experimental data This is from michael the side lab with east and this is kind of free analysis for the long-term experiment by richard richard lansky and what is Data shows is sometimes you have situations where you start in a simple vanishing environment and then Some ecological diversity emerges then it persists for a long time and then it collapses again So this is in this case is this is just two species species diversity But it kind of is reminiscent of this Intermittency kind of in a much simplified context. So I'm hopeful that as we Uh Otherwise As we have more longer and longer time series data and richer data will start seeing this phenomenon More okay, the final mechanism I wanted to point out which is very interesting Is sometimes you see these long persistent communities. They can be more than five species six seven species that persists seemingly indefinitely They look like an evolutionist table state, but then you stop the mutation and then disappear So how uh, that might be even possible So this state here multi species state persists despite a barrage Of mutations, but then you turn off all the mutations and everything collapses. Well, it turns out That uh What stabilizes communities are a loss of function mutations which we kind of put into the system because we thought oh This is something that will happen biologically. So like producers sometimes will lose their production and go directly to being resistance Or the greater will become sensitive or producer will become sensitive all these loss of function mutations the If if the corresponding source strains are abundant The they'll kind of add almost deterministic correction to the ecological dynamics and these dynamics turns out to be stabilizing Now I want to emphasize that this is not neutral Neutrally maintained community because you have coexistence of strains that cannot be recreated by these loss of function mutations okay, but But their stability is maintained by the loss of function mutations. So we have this strange persistence without ecological stability Okay, and then we have some other mouths And kind of the final thing. I just wanted to to mention and I'll be very quick is That all these things that I told you the emergence of a revolutionist table state All these interesting Equal evil mouths. They only work In the finite mutation Limit, okay If I if I tune down the mutation rate alone Everything interesting in this model disappears And what I mean by mu going to zero limit is that the limit where I introduce a change to the community Wait a very long time went wait for the community to equilibrate And only then introduce the next mutations to the community then wait for a very long time Uh, okay, and we know that in practice in for pretty much for every community It won't be the case right you have many mutations arising in the community So once the mutation arises and kind of leads to some ecological transient another mutation will also arise so this is kind of The realistic limit here But kind of the strange thing is just almost always when we think about Evolution dynamics very implicitly in this one change at a time and let's wait and see what happens Kind of mindset at least this is how my thinking was okay So this is the reality Finite mutation rate we see all these different things in this Mu going to zero limit where mutations happen one at a time and we'll wait long interval between them The model is completely boring Uh, and there's this disconnect between reality and how we think and really this captures The essence of equal evolution dynamics, right? Ecology and evolution happen on the same time scales if you bring the evolution in timescale mutation rate way below what is reasonable Uh, things become boring. So this is an example of a model which is only interesting in the equal evolution limit and then becomes boring in the In the kind of in the kind of the traditional limit we have and just to kind of Make one final analogy and then i'm done Uh, what I was arguing on the slide that skipped that you can Make this a good analogy is between mutation rate and temperature. Okay. The mutation is what creates these fluctuations in the system Now in physics, we talk about these states of matter. They're like, uh, liquids gases crystals Liquid crystals plasma glasses you name it many many different states of matter But many of these states of matter. Let's say liquid. They only make sense At finite temperature there is no such thing as liquid That's uh in the zero temperature limit. I mean there are superfluids, but that's a separate state of matter, okay So it's fundamentally impossible to think of liquids without fluctuations Similarly all these interesting things that I showed you they don't exist in the small mutation limit But that's because these evolution fluctuations are fundamental to their existence and since this is The realistic situation for the entire ecosystem Mutations are not too rare. I think that's where we need to focus on To to study equal evolution dynamics Okay, and I'll just summarize we found Emergence of diverse communities beyond that we've identified all these interesting equal modes That suggests that we need to go beyond classical notions of ecological stability evolution stability This one mutation at a time mindset when we think about evolution Now data sets long-term data sets are still kind of There aren't many very many of them. It's hard to do Experimental evolution, but one thing we can do right now that we are ready for Is we have these powerful computers we can do this Large-scale multi-scale equal evolution simulations and And we can do a lot more of that in many different settings. Look at these equal evil modes and see If if there is an infinite diversity of such modes or there are only like few that keep repeating Okay, so thank you