 what my lab is working on these days. It's been really a wonderful conference, chatting with all of you after the talks, listening to talks and so forth. I think there are a lot of overlapping interests, certainly between my group and many of you. Today, what I want to do is tell you about ongoing work, so I'm not going to show you any published work, meaning that in one slide, I'll tell you about what my group has been doing in the past, just so you know what you can come and chat with me about in the next couple of days. So as Simon mentioned, one of the major interests in my group has been studying sudden transitions in populations. So in general, we're interested in how interactions within a population or a community can lead to interesting evolutionary and ecological dynamics. So one of the most basic things that can happen is that if you have cooperative growth, you can have a sudden transition in response to a deteriorating environment. And we've been interested in early warning indicators, the universal behaviors of populations near these tipping points. Now, any time that you have these sorts of cooperative behaviors, you might also want to ask, is it possible for sort of non-productive cheats to emerge and spread in the population? So we've been interested in these sorts of dynamics. And in particular, what sort of ecological conditions might favor or disfavor the emergence of these cooperative or cheating-type behaviors? And some of this work was actually done by Al Sanchez, who's now at Yale. Well, not currently at Yale. Right now, he's here, but in general, he's at Yale. So much of this behavior is here where we're looking at cases where you have secretion of some sort of enzyme that helps to break down a complex sugar source. So this is what you might think of as the creation of a public good. We've also been interested in an analogous set of situations where it's the degradation of a public bad. So there are many contexts in which antibiotics, the way in which a population gains resistance, is via degradation of the antibiotic. So if you have degradation of some antibiotic, then you may be able to protect sensitive cells. And so we, for example, see coexistence between resistant and sensitive populations, for example, in ampicillin. Moreover, you can also, well, we've also demonstrated that you can have a mutualism. So you can have two antibiotic-resistant strains that together can grow in the presence of both antibiotics because they're each breaking down one of the antibiotics. And then finally, we're interested in a variety of kind of spatial dynamics. How is it that cooperative growth processes will alter the kind of spatial patterns that you get? Much of this work has been in collaboration with Kirill Korolev, who was also a postdoc at MIT and is now a professor at BU, also here. So you guys should chat with them about those projects. Kirill's also involved in a number of the early warning indicator projects. So these projects, I think, highlight the range of the interactions that you can get when you have interactions within a population. But in almost all of that work, it was really interactions within a single population, within a single species. And of course, the problem that we're all facing now is that that's not the way that these microbial populations are behaving in nature. They're in these complex communities containing many, many species. This is too complicated for me. So we are kind of, a lot of our work now is focused on the multi-species communities, but when I say multi-species, I mean n greater than one. So we're kind of taking a bottom-up approach to this problem where we're basically trying to understand pairs, trios, and kind of marching our way up. It's not obvious to the degree to which the insights that we get in these simple communities are necessarily going to hold in the complex communities that we all care about, but you have to start somewhere. And so this is our approach. All right, so I'll tell you a few different kind of projects in my group. All right, so first, we've been recently thinking about what are the typical ways in which communities will change in response to a deteriorating environment, in particular in response to increased mortality. Just very generally, it seems that increased mortality favors fast-growing species. This is true in pairs as well as within simple communities. All right, so we've also been interested in these kind of network of pair-wise competitive interactions. So we found that they're surprisingly informative regarding the structure of multi-species communities. And so then there's the question, well, what sort of network motifs or structures that in natural communities may help facilitate the diversity that we see in nature? And one of the proposals is that there could be a lot of these non-transitive or rock-paper-scissors type interactions. And what we've found is that, at least in the communities that we have looked at, so we can't find any of them. So we've looked at thousands of trios, can't find them. So our sense is that these are probably not common in natural populations, although this is something that everyone likes to argue about. And then finally, on a slightly more mechanistic side, so much of the approach that we take in my group is sort of phenomenological in the sense that we are trying to characterize the interaction using some simple parameters and asking, to what degree can we predict what happens later? But it's also good to go and look in and say, well, what is actually mediating the interaction? And one thing that we've found that is surprisingly dominant in determining the structure of the communities that we've been looking at is modifications of the pH of the environment. So just to get started, so this is a project that was being led by Clara Breu, who is a physics PhD student in my group. And what she wanted to just ask is, well, if we were to impose a general mortality onto a community, what is it that you might be able to predict in general? So the idea is that we have some complex community and it's gonna be put under stress, or this stress could take a variety of forms, antibiotics, diarrhea, attacked by the immune system. So some of these could be specific, but right now we're gonna try to imagine a general increase in mortality. So what we wanna know is, what is gonna be the structure of the resulting community? So as I said, we wanna start simple, so we're gonna start with pairs. And in particular, I think it's worth highlighting. In our group, we've been studying a lot of pair-wise competitions, and almost all the time what we see are simple outcomes. So there's basically three things that we see over and over again. It's not just, and other people see these things too, but I just wanna, it's good to be clear about what the simple and that, because it's easy to imagine that everything has to be infinitely complex, but what we find is that at least in the context of pairs, things tend to be simple. So one thing that we often see is dominance, where if you just compete two species and you measure the fraction over time, what you see is that one species outcompetes the other species. So dominance and competitive exclusion. Another thing that we often see is coexistence. So you compete a pair of species and they coexist in that environment. At some equilibrium fraction that's independent of the starting fraction that you do the competition at. And then finally, we sometimes also see bistability, where the outcome of the competition depends upon how you start. So what you see is that if you start with a lot of one species that's gonna win, and vice versa. And this is actually all experimental data, but just characterizing the three typical things that we see when we compete pairs of species. So maybe 95% of the time we see one of these outcomes, 5% of the time we see something crazy. Crazy could mean a variety of different things. Sometimes there's just rapid evolutionary dynamics, so you mix and you see that things kind of go weird. It happens, evolution, you can't always forget about. Sometimes we've actually also seen bistability with coexistence in one of the states. So here, this is bistability with exclusion on each side, extinction in one or the other. We've also seen a few cases where it's bistable, but one of the two states has coexistence. And then finally, occasionally we've seen oscillations, well, at least once in the context of this mutualism between the antibiotic resistance strains, we actually see period three oscillations. So these sorts of things can happen, but in almost all cases we see one of the three things on top, which is convenient because those are the basic outcomes that arise in just the simplest models that you might write down of pairwise competition, in particular this competitive lotcavoltaire model. So the idea here is that you just have two species that are inhibiting each other's growth, and depending as a function of the strength of the inhibition, then you get these three, you could think of it as four outcomes, here there's dominance and then exclusion, kind of the same thing, flip side, coexistence and bistability. So you get coexistence if there's weak suppression of each other's growth, and you get bistability if there's strong suppression of each other's growth. Just to be concrete, and I'm gonna put in, this is kind of the only equation I'm gonna include, just to be concrete, so this is the two species, this is the per capita growth rate here, and what it's determined by is just a few things. There's a maximum growth rate, this R1, it inhibits itself, in this case the assumption is it inhibits itself logistically, so it just decreases linearly like this, and then finally there's the inhibition by the competitor. It's a very simple model, and I wanna stress that this model is certainly not true. In any actual set of species, none of the assumptions in here is actually what's happening. So the question in my mind is, what are the qualitative predictions of a model, and do they actually pan out in a given experimental system? So in particular, what I'm gonna do is just ask, if we take this model and we add a uniform mortality, what happens? So uniform mortality means we're just up here and we just add this delta term. So this is just meaning that both species are experiencing some increase in mortality. Very simple assumption. It turns out in the context of this model there's a very simple prediction because this delta can actually just be brought into those other terms, and so you can just redefine the parameters within this model, and there's a simple geometric interpretation. So now instead of N1 and N2, I'm calling it NF and NS, we're assuming that one of them is gonna be faster, more fast growing than the other, so RF is larger than RS. So in general, that's just a label so that you can assume that without loss of generality. And what happens in this context is there's something very simple, that as you increase the mortality rate, the prediction in this model is that you should just travel down and to the right at a 45 degree angle. So for example, if you start up here where the slow grower is winning, so it's slow growing, but maybe a strong competitor, so it could still win, as you increase the mortality, the expectation is that you'll travel down this 45 degree angle where you're favoring the fast growers. Eventually, you expect to make the fast grower win, but along the way, you expect to cross either a region of coexistence or a region of bi-stability, but not both. So again, this model is certainly not gonna be an accurate or quantitative description of any given real pair of species, but this is a qualitative prediction of the model that could be robust to the fact that the assumptions are not actually gonna be true in any given system. But that's an experimental question, right? All right, so yeah, so then what we did is we went and we do a lot of competition between species, so we basically just started looking at some of our species where in particular, the slow growers seem to be strong competitors. And what we found is that this basic picture is surprisingly predictive regarding what happens in pair-wise competitions. And important, yeah, so let me do it. So I'll tell you about the experiments. So what we're doing is we're just competing pairs of species in liquid, all right, and we allow the populations to grow up. And then each day we're gonna perform a dilution where we transfer a small fraction of the cells into fresh media, and then the population grows back up or the community competition occurs again. We just repeat this process five or six times, typically 50 to 70 generations, so that we allow the competition to reach resolution as in the data that I showed you before. Okay. And in particular, what we're gonna do is we're just gonna increase this dilution in order to impose and increase mortality on both species, right? And indeed, it's symmetric, it's imposed on both species. And this basic picture works surprisingly well. So what we see, for example, is that when we compete pseudomonas veroniai against enterobacter aerogenes, what we see is that PV, the slow growers, winning at low dilution factor, but then as we increase the dilution, we cross this regime as kind of expectants. I'll show you the data, but just give you a picture that this is what we think is happening, right? So first of all, if we look at the fast grower fraction, that at low dilution factor, it goes down, right? So we get exclusion of the fast grower, that's fine. Question is, what happens as we increase the dilution rate? And it's really just as we were kind of expecting, which is that you see that the fast grower, it loses at low dilution factor, we get coexistence at intermediate dilution factor, and then we get the fast grower winning at high dilution factor. I'm sorry that it's very hard to see those lines, but yes. And the way that we like to visualize this sort of data is via these things we're calling subway plots, where what we show here is basically the color of the species is telling you about whether that species can survive at equilibrium when in competition at that dilution factor. So here, blue is saying that only that slow grower is surviving at intermediate dilution factor, we get coexistence, and at high dilution factor, we have again competitive exclusion where the fast grower is excluding the slow grower. Yes, yes, so indeed in all of these conditions, the monocultures can survive in all those conditions, yeah. But I think what you're highlighting though is that at some limit, this thing has to be true, right? Because if you dilute faster than the slow grower grows, then it's gonna go extinct and the other one should win. So indeed that in the limit, it's kind of trivial or it has to happen, but it happens in very particular ways, I guess. Yes? Okay, so the question is about the slow grower, like why is it? Yeah, yeah, right, so we've been characterizing many pairs just to get a statistical sense of what's going on, so we have not in this case delved into what is the mechanism behind that. Okay, so for example, it could be that the slow grower is making a toxin that is costly to make. Yes, sorry, sorry, slow grower, yeah, so yeah, exactly. What I'm referring to is the growth rate on its own, yeah. And it's the growth rate at low density on its own, yeah. By sampling both species. Oh, I see, I see, because by sampling, you're imposing immortality, is that, yeah, yeah. And if you want to do that, that's fine, but you have to be aware that you are doing it, right? And I guess I'm a little bit embarrassed by some of this stuff because we've been spending years doing competitions or, well, at least thinking about this stuff, and we often, I didn't think too hard about what delusions we were using in the sense that we would often just say, oh, yeah, this pseudomonade, it out-competes this other species in minimal media with x-me-o-e-o, so we define the media, but we often don't feel we need to specify the dilution factor, yet what you can see is that I can make anything happen by changing the dilution factor, right? So, you know, so it's, I guess, a warning to my group and also, you know, to all of us, yes. I mean, we're only allowing a fraction of each of the two species to survive when we do the dilution. So in that sense, I would say it's the same mortality that is, same mean mortality is experienced by both species. Yeah. Oh, I mean, we spend a lot of our life thinking about density-dependent effects, for sure. Oh, no, I mean, for sure, and again, that model is super simple and none of the terms in the model are correct, and that's what I find surprising is that the qualitative prediction of the model works not just for this species, but for others, despite the fact that life is complicated, yet, yeah, I guess, you know, I'm always looking for what simple things can we say despite the complexity of life, right? And I guess this seems to be a pretty robust thing, right? That you favor the fast grower and in between, you cross either a region of coexistence or a region of vice-tability. All right, so maybe I'll, okay, so I showed you data for one pair, but we've done this for many pairs. And in particular here, what we've done is we've taken four different species with different growth rates. What you can see is that at low dilution factor, the slow growers do better than the fast growers, right? But as we increase the dilution factor, we can plot just the mean fraction of each of those species in all the pair-wise competitions, and you see there's just very clear patterns as we shift. What you see is that the slow growers do worse and worse, the fast growers do better and better, and then you can kind of interpolate between them, right? Yeah, indeed, okay, yeah, so in almost all these cases, there was only one pair that had vice-tability. So in all these cases, basically what we did is we started with a bunch of different fractions. And in only one of these pairs was their vice-tability, so it actually doesn't, it doesn't, this sort of data is not affected very much by that, but yeah, we started with a wide range of fractions. And all right, so this is how the pair-wise competitive outcomes are varied as we vary mortality, right? Favors the fast grower in stereotypical ways. And the question is, well, what can we say about more complex communities? Well, so luckily we've spent a fair amount of time over the last few years thinking about these community assembly rules. So based on pair-wise competitive outcomes, what can you say about simple communities, right? So first of all, this is the subway plots for three other pairs. What you see is that the two fast growers, they coexist at all dilution factors, and that's because their growth rates are so similar, you don't have much of a lever arm as you vary that dilution rate. So then you don't change the outcome. But the other two pairs, they have a difference in growth rate, and again, you have the slow grower waiting over here, a region of coexistence, and then the fast grower wins. And the question is, all right, so these are three pair-wise outcomes. What can we say about what would happen if you have all three species together? And in particular, we have proposed these community assembly rules that are based on qualitative pair-wise outcomes to qualitative trio outcomes or more. And basically, what the proposal is is that we're gonna say that a species will survive in the multi-species competition if and only if it's not excluded in any of the pair-wise competitions. In particular, if you look down here, for example, the yellow species is out-competing each of the other two in pairs, so then we would predict that it's gonna out-compete them in the context of the trio, so only maybe yellow will survive down here, and we can kind of apply this rule all the way down. And at least, for example, in this trio, that assembly rule works for all the dilution factors. What happens is that the yellow is the only one that survives at low dilution factor. Over here, at 10 to the five, you see that there's coexistence of all three pairs, so you would then predict coexistence of the trio, and indeed, that's what we observe experimentally. And then finally, at high dilution factor, since the yellow species is out-competed in the pair-wise, you would predict it to be out-competed in the trio, and indeed it is. As before, what we found is that these assembly rules predicts, correctly predicts survival about 90% of the time in the context of going from pairs to trios or four or five, and indeed, we see, again, similar things. You hear it, it happened to work for all of these dilution factors, but we've done all the trios in the four. So indeed, 90% of the time, we correctly predict survival. So it's not 100%, and you wouldn't expect it to be 100%, but it's, I'd say, not bad. And just to be concrete, in the previous experiment, instead of varying the environment, what we had done is we had just taken eight species and done all the pairs and trios in a single environment, and that's kind of where we had first developed these assembly rules, and so that generates some sort of network. Oh, and incidentally, we've also recently found that these assembly rules work surprisingly well in determining the structure of the gut microbiome in the worm. In particular, if you apply these rules, you can predict the fraction at equilibrium in the gut of the worm within about 10%. Actually, so the median across 20-some trios was an error of 10%, which, frankly, is close to our measurement area. I mean, if you just do the same thing. So it's surprisingly good, I would say. So given that these pairwise outcomes seem to be so helpful in sort of guiding our thinking in the context of determining community structure, a natural question is to ask, what does this network look like if you were to sample it from some sort of natural community, some co-occurring set of bacteria? Because all of the species so far that we've been studying were just from random soil isolates taken from different parts of the world, and then we just put them together and see what happens. So to get at this question of what it is that this network might look like within an actual community, what my microbiology student Logan did is she just, well, she sampled from a single grain of soil, isolated 20 species, and then measured all the pairwise outcomes. Now, the reason that she wanted to do this is because one of the major questions in ecology is to try to understand how it is that the diversity is, that we see in natural communities, how is it that that's possibly stable? And I'm not gonna solve that question here, but this is what we're trying to figure out. Yeah, yeah, so I went out to the front yard, dug 10, 70 years. So oxy, Albee, that sort of thing? Oh, yeah, so all of this is done in the presence of oxygen. We have a chamber that, yeah, but this is all done in the presence of oxygen. So maybe the structure of this network might give us some insight into how it is that these communities can be so diverse. So the question is if you go and you look at some set of species and you measure their competitions, how they interact, then maybe you might see some patterns in that network that would provide guidance in terms of what's going on. So you might, for example, see something like modularity where different subnetworks are strongly interacting or maybe you'll look up here and you'll see something like these non-transitive rock, paper, scissors type interactions. And the reason you might expect those is because they're predicted to increase the diversity in the community because if you have something that looks like this, then no species can get to abundant because that opens the door for another species to be spreading. So there's a great deal of theory in this field basically asking, as competition goes from being very transitive, like a competitive hierarchy, like you were talking about, to more non-transitive, then you expect to be able to get more diversity at equilibrium. But this is an experimental question. We have to go out and we have to make measurements to see what these networks actually look like. So what Logan did is she just walked out of the door of the lab, she went into the grass, she dug 10 centimeters down, grabbed a grain of soil, brought it back to the lab, vortexed, plated in the presence of oxygen. And I said many species, but in particular she chose 20 of them to study further and do all the pair-wise competitions. And just to tell you, yeah, so what we see is basically what Karina was talking about before which is we found a very strong competitive hierarchy. So it's really that one species with basically the best could beat everybody, this species could beat everyone except for the first, all the way down. So this is the kind of network that should be least favorable for leading to diversity, but that's what we see. I'll show you the data just so that you can believe that we did a lot of work. This is the 20 species arranged in order of mean competitive fraction in all of the measurements. And what you can see is that basically there's a lot of red down here meaning that these species are outcompeted, go extinct as a result of competing with the vertical one. A lot of green up here meaning that these species drive these other ones extinct. Blue corresponds to coexistence between the two and yellow corresponds to bi-stability. So in all these pairs we did a 5%, 95% starting condition so we could assay for bi-stability. And what you see is that there's a very strong competitive hierarchy. Yeah, so this is measuring after 60 generations, what happened? Yeah, okay, so this is a mixture of the alphas and the Rs. This was all doing daily dilutions of 100. We thought that maybe this competitive hierarchy was driven by the dilutions, so then we went and we repeated this for a low dilution and it was still hierarchical. So yeah. Yes, except that once you add a dilution in there then it couples to the Rs like what we were talking about before. Yeah, so there are many possible explanations. This is one of them. Yeah, I'm not gonna give the answer for why these things coexist because I don't know. But it was Roy first, yeah. So the question is how many of these competitions can be turned around by doing different dilutions? I would say that over the range that we have access to I think we could probably alter a third of them. But we've only done two. I mean it's a lot of work to do one of these matrices. Yeah. Yeah, yeah, so this was one 10th LB. We've also done it in a defined media. We've also seen hierarchy. So yeah, of course, this is not the soil in a lot of different ways, but as far as we can tell we have not been able to get rid of the hierarchy despite a fair amount of effort to do so. In the sense of different medias, different dilutions. Yeah. So okay, so this is, okay, so the question is whether the logical Volterra model predicts oscillations. So unfortunately not this logical Volterra model. It's the predator prey logical Volterra model. Yeah, but we basically don't see oscillations anyway. So yeah, yeah in the back. Yeah, so yeah, all of the data that I've shown so far the species abundances were determined by plating because they have different colony morphologies. About the genus, yeah. Oh, the cheaters. Oh, okay, so yeah, so there could be many complications here. Yeah, so all of the competitions we've done here were done from single colonies. So we don't, so this is, we're trying to study ecology rather than evolution, right? And as I said, a few percent of the time you can't get away from evolution, but by and large these are clonal populations within each species and we're competing species. Yeah, so we're really just trying to focus on the ecology. Extreme competitive hierarchy and in particular you can go and you can look at all the different trios and ask how many different cases of this sort of rock, paper, scissors type interactions can you find, right? So with 20 species there's over a thousand trios that you can look at. So we went and asked, okay, can we find any rock, paper, scissors? And the answer is no, right? So basically there were no robust examples of rock, paper, scissors among these 20 species. So our sense is that, well we spent a lot of effort trying to find these non-transitive type interactions and we can't. So my provisional takeaway is that they're not playing a significant role in stabilizing the diversity we see in natural communities. So there we are. Okay, so I'll probably not finish what I was going to say, but it's fine, all right. Okay, all right, then I'll try to be clear. So that you don't have burning questions. All right, so far what we've been doing is taking this purely phenomenological approach. We're just asking what is the interactions, but there's some effective interaction between the species. What can we say based on that? And in particular we were imagining that there was some direct interaction between the species influencing each other. But of course in the context of microbes most of the interactions are mediated through the environment. So things are being secreted or uptaken from the environment and that's influencing the growth of other species. So the question is what can we say about this sort of environment mediated type interaction? In particular my postdoc, Christoph has found lots of examples where it's really the pH that's determining an awful lot of what you might want to know. So first of all, microbes when they grow they change the pH of the media. I think we've all seen this, but I just want to highlight that it's ubiquitous. This is just, we took 119 different bacterial species that we'd gotten from the soil. We grew it up in some vaguely reasonable media. We measured the pH afterwards and what you see is pH goes everywhere. We started pH seven, there's a couple species that stay there, but almost all of them changed pH. Some of them acidify, some of them alkalize, a variety of effects associated with the growth in terms of the pH. The question is, is this important? And we're gonna argue that the answer is yes. So what we see is that the pH can lead to all the kinds of phenomena that we've been studying in the context of a lot of these other interactions in the context of, for example, breaking down sugars. So first of all, you can have positive interactions where microbial species is, say, acidifying the environment and that's good for it. For example, budding yeast does this, right? And this leads to all the stuff about cooperative growth, minimal viable density required for survival, all the stuff that we and others have been thinking about in the context of cooperative growth shows up for pH as well. We also see that there are many cases where microbes change the pH of the environment in a way that's bad for it and this leads to something that you might call ecological suicide where a population kind of kills itself. And then we also see that just by thinking about the pH, we can often get a good prediction regarding the kinds of outcomes that might happen when we compete pairs of species. So some people think that Easter Island population of humans had this ecological suicide kind of thing where Easter Island started looking something like this and then humans came, they chopped down all the trees and then they left behind this. There's a lot of arguments among anthropologists about whether the human population on Easter Island actually did this. I'm not gonna try to convince you that that's the case but I will try to convince you. And incidentally, In Yu Huang on Monday talked about a quorum sensing mutant that did really precisely this thing here where in that case it was alkalization but what we see is that many different species will change the pH in the environment in either direction in a way that is so bad for it that it kills itself. So this is just, this is an example of data for a particular species where we at high kind of temporal resolution went in, plated to measure the CFUs. What you see is that the number of live bacteria, it grows exponentially but then very rapidly dies exponentially and after 24 hours and we actually, there are no viable cells left in the media. And indeed, this is accompanied by a acidification in this case. Doesn't prove that it's pH but we can, for example, increase the buffering and what you see is that eventually you can save the population by adding a sufficient buffer. So it's not, again, this is not just this species surprisingly common actually. And given that this is already kind of a weird phenomenon, there are a bunch of corollaries of it that are very natural. In particular, what we can do is we can harm the bacterial population in order to save it. So what we can do, for example, is do a bunch of different things that harm it. We've done antibiotics, ethanol, salt, lots of different things to harm the population can actually save it. So what I'm showing you here, for example, is the fold growth experienced by the population as a function of an antibiotic that we're adding. In the absence of the antibiotic, the population's all dead. It's this ecological suicide phenomenon I was telling you about. If you add enough antibiotic, you can kill the population again. Not a huge surprise. But there's this region of antibiotic concentration where you can save the population because you're inhibiting growth and preventing that acidification that was going to kill the population. I think that this is interesting consequences if you think about what happens when you have mixed populations, communities, resistant sensitive, I'm not gonna get into it, but it's a kind of a natural surprising consequence. Yeah, so this is just batch growth over one cycle but we've also done multiple growth dilution cycles and the same basic thing happens. And it's common. But I do wanna just highlight that this interactions between the pH, there's just a few different kinds of interactions that you can imagine. Basically, because there's different combinations of whether each species is acidifying or alkalizing and whether that's good or bad for the species. So what you can do is you can just make a little table where you can say, well, what is it that the pH that they're imposing on the environment and what is it that they prefer? So you get cooperation on this diagonal and you have inhibition, possible ecological suicide in the other one and then you can ask, all right, what are the possible outcomes that can occur based on combining these different species? And there's some very, I think just basic interesting qualitative outcomes. So first, if each species is changing the pH in a way that's good for it but bad for its competitor, you might imagine that could lead to by stability between the population and indeed that's something that we can observe experimentally. You can also imagine that there's a case where species is trying to change, is trying to say acidify the environment and make it, and it's good for it but can't quite do it on its own but could do it with the help of another species that maybe you could get an example of succession where a second species helps the first species survive but doesn't survive at the end of the day and indeed we see that kind of, that outcome experimentally as well. You might also imagine that there could be something that you could call a murder suicide where it's an environment where a given species is gonna be killing itself and along the way it kills its partner species and indeed we see that experimentally. And then finally you can imagine that if each species is doing something that's bad for it then if, but in opposite directions then they could stabilize each other, right? So there can be an example of a mutualism and indeed we can see that experimentally as well. So this is all just based on knowing how the single species behaves in a monoculture and how the pH is altering its growth, okay? And I'll do the last slide because I think it's fun. So in this case of bi-stability that I mentioned where the two species are each changing the pH in a way that's good for it but bad for the competitor, you can think about this in the context of alternative stable states. In the human gut microbiome there's a lot of attention that's being paid to what we think are alternative stable states. Between for example the C. diff dominated unhealthy state with the healthy gut microbiome and what we would like is to know how to transition between these alternative stable states, right? So we've been using this as kind of just a model simple little community that we can use to ask well what is it that we can do to perturb this community and induce transitions between them, right? So we found a variety of different kinds of abiotic perturbations, right? So different perturbations that we impose but there's also something that we found that I think is really kind of surprising and fun which is that we can induce transitions between these alternative stable states via a third species that acts as a transient invader. So what it does is it catalyzes the transition from for example the LP, the lactobacillus dominated state to the CA, right? And what's interesting is that it doesn't survive the transition, right? So it can induce the transition but it can't survive that final state. And interestingly there's a fourth species that we can add to induce the transition in the other direction where we can again make it go back to this state and again this species doesn't survive, right? So there's, well there are many cases where we observe these transient invaders where they can induce the transition but there's no evidence that that's what happens. Later you look at it in the community just switched and there's no evidence that that species was ever present and it makes you wonder to what degree species invasions could induce transitions for example in the human gut without showing leaving any trace. And I don't know if that is real or not but at least here we seem to see it. All right and so with that I'll summarize. All right so we've been interested in this bottom-up approach to try and understand multi-species community assembly. So here we found that there's just some very simple things that you might expect will happen as you increase mortality in pairs and in simple communities and that seems to be borne out in our experiments. We'd love to be able to see this kind of thing in for example the human gut with diarrhea. I was talking to Tammy Lieberman actually last week about whether you could see these kinds of things in complex communities like the human gut or in the worm. We've also been interested in these sorts of network structures and one of the dominant ones that people talk about are rock, paper, scissors trios. There are experimental examples of these but my sense is that they're not actually so common in natural communities. And finally we found that the pH seems to be a dominant determinant of how a lot of these species are interacting in a way that, I mean of course there are many examples where people look at the pH in the context of the oral microbiome and others where it seems to be structuring things and it's just a matter of trying to figure out which, what are the list of things that you should first look at when trying to determine the origin of the outcomes that we're seeing in various systems. And with that I will thank the group who are the ones that have actually been doing the work. You guys should all introduce yourselves to Carol and Al who are here and I'll also maybe just mention that I have three postdocs that are gonna be starting groups this coming fall. So if you guys are looking for postdocs and so forth you should check them out. So Nick Vega is starting a faculty position with Bruce over at Emory. Jonathan Friedman is starting the Hebrew University of Jerusalem and Alfonso is gonna be starting a group in France with a CNRS position in Toulouse. So yes, thank you.