 Just wait for all the participants to come back from the breakout rooms, and we'll start. OK, I think everybody's back. So it's my great pleasure to introduce the next lecturer, Alvaro Sanchez. Alvaro is a professor at the Department of Ecology and Evolutionary Biology at Yale University. And this work focuses on predicting microbial dynamics and microbial evolution. And in today's, in actually this cycle of lectures, we will talk about the assembly and the evolution of microbial community. And today is giving the first of three lectures. So thank you very much, Alvaro, for being with us. And please share your screen when you are ready. Can you hear me OK? Yes, perfect. All right, wonderful. Thank you, everybody, for taking the time to be here today. So I'm going to give three lectures on the topic of microbial community assembly. And the three are very interconnected. So and I'm thinking of them as very, I think, relaxed. And the idea is that you can get, you know, have three hours to explain how at least I see this problem. And I'm hopeful that if you have a question or anything, you can interrupt or let me know at any point. So as I said, we work on these very fascinating problems of understanding how microbial communities form and how they evolve. And I think this is a very important problem for many reasons. But perhaps the main one is because microbes are everywhere. They have colonized all of the sources of the earth. You find them in the bodies of all animals and plants. You find them in the soils and the oceans and absolutely everywhere. And in all of this habitat, they carry fundamental and critical functions. They are responsible for the vast majority of nitrogen fixation, for instance, in the atmosphere. They directly impact the health of their hosts with animal and plants. And for that reason, understanding the composition of these communities is so important. And for those of you who are not necessarily having paying attention to the entire field of microbial ecology, I'm sure most of you have heard about the human microbiome. This is the realization that has been more acute in the past couple of decades, that we share our bodies with an enormous number of microorganisms that have colonized basically all of the surfaces of our bodies that are connected with the outside world. And microbes in our skin, in our gut, in our mouth, our stomach, and essentially anywhere where they could colonize from the outside world they have. And in all of these areas of our bodies, they form these very dense and complex ecological communities containing a large number of species that interact with each other in really interesting ways. And the outcome of these interactions are the assembly of an ecological community, of microbial community, which, as I said before, can have important effects for our health. And given the importance of these communities, a critical question is whether we can predict the assembly of these communities in a way that will allow us to manipulate them and to steer these communities towards states that are favorable and away from those that are not. And this is a question that is very close to the heart of my lab. This is one of the ones that I think we're interested in. And if we want to develop a predictive theory of microbial community assembly, one that can tell you, OK, if I manipulate this, if I can intervene in a microwave in this way, I'm going to observe that response. If we want to walk towards having such a theory, there is even more fundamental question that we need to understand first, which is how reproducible microbial community assembly is. And the reason for this is that the only way you can predict something is if it is reproducible. For instance, imagine that you did an experiment 10 times, and every time you got a different result. There wouldn't be anything you could predict. By the other hand, if you could do the same example experiment multiple times, and every time you get the same result, and there's some signal, there's some phenomenon that you can aspire to predict. So that's a very fundamental question, which is how predictable is the process itself? And as I will discuss in the next two lectures, the answer is very interesting because it depends on the level of organization at which you're looking at it. And I wanted to start by giving you a specific example and a paper that I really liked that came out a few years ago by Stenos Loka and Michael Dovely. And this is a team of researchers that were studying the assembly of the microbial communities that form within the foliage of bromeliad plants. So as you can see here, at the base of the foliage, bromeliads have a little hole, a cavity that fills up with water. This water is steaming with microorganisms. We have bacteria, archaea. You have protists. You have all kinds of other microbes. And it is believed that the composition of those communities can be very important for the plant as it helps them fix nitrogen and some other things. So this is also a very interesting model system for understanding this question of how reproducible community assembly is because one could go to the same geographical location. These plants, I believe this study in particular was in Brazil, bromeliads are a tropical plant. And you can look at a set of plants that are in close proximity to one another and examine the communities that are assembled in them. And these are plants that are said, you know, the habitats in which those communities form are very similar. The plants are from the same species. They're experiencing very similar temperature, humidity, and potentially also you should be they should be colonized, but very similar bacteria that are in their environment. So it's a very nice natural system to investigate how reproducible microbial community assembly is. And just because I wanted to walk you through what kind of data they collected and because the data that we collect ourselves look very similar, I wanted just to take some time to explain to you what they observed. They looked at what? I think a few dozen plants that were in close proximity to one another, and they measure the microbiome on each one of those plants. And the data looks more or less like this. This is the proportion of the microbiome that is made up by different taxa, this operational taxonomic unit, which is you can think of this as the lowest level of resolution, like species level, for instance. And they're looking at the composition of the microbiome of one particular plant. And each color here represents a different OTU and the width of this color represents the abundance of that OTU of that species in the community, right? So if, for instance, this green or two color here, the width is very small, means that the abundance is small, whereas this brown color here means that the width is larger, it means that that species was more abundant. And when they looked at the composition of plants that the communities associated with different plants, they found that they were very variable. Less than 1% of all the species were found in all of the plants. And as you can see here, each color represents a different plant and most of the plants are contained a very different, they're a different community. Now, intriguingly, what they found, however, is that when instead of looking at the taxonomic composition of the plants like they did here, they looked at the metagenome of its community. They look at the abundance or the abundance of genes involved in different metabolic pathways. They found that if you do that, the result for a given plant looks something like this, right? In brown here, they're looking at the abundance of genes involved in fermentation, in blue genes involved in oxygen respiration, carbon fixation, and a bunch of other metabolic functions. And when you look at that level of resolution, community assembly was much more predictable. And when you look at the metagenome of different plants, you find that they're quite convergent and similar to one another. And this idea, very similar results have been reported in other systems as well. For instance, in studies of the microbial communities associated to microalgae, the second found that community assembly, when you looked at it at the level of function and the metagenome, they found communities that were very similar to one another when they look at different individuals, yet when they look at the taxonomic composition, it was much more variable. And the same has been observed, for instance, for various body parts in the human microbiome. Again, taxonomic composition even at the phylum level could be quite variable for different subjects in the same body part. Yet when you look across different subjects, but look at the level of metabolic pathways that you find the different fractions of the metagenome that are devoted to different metabolic functions, you find that different body parts have distinct microbiomes and that they are much more similar from host to host. There's still variation, but it's much less. And that has been proposed as little as the proposition that of a paradigm for the organization of microalgae communities. And this was again by Astieno's local colleagues where they proposed that similar environments should promote similar microbial committee function while allowing for taxonomic variation within individual functional groups. So you could have, they're within a functional group, you could have that different taxa can differ in different habitats can differ in which taxa they contain. Yet when you look at them at the level of function, they are much more similar to one another. Right, so this is pointing to the existence of a generic organization of micro communities. And this question that I posed at the beginning of how reproducing micro communities assembly is, then it seems like the answer will depend on what level of organization you are studying. So, but even though there seems to be a rule that emerges when you look at a large number of different habitats, it's still a very clear where it comes from, right? And what is the origin of this principle, right? That seems to be found so often. So to understand this better, it's worth thinking a little bit about the forces that shape the assembly of microbial communities, of ecological communities in general, right? So ecological communities are the outcome of both deterministic and stochastic processes. So on the deterministic side, you have selection, right? And here the idea is that in a given habitat, you're going to have some taxa that grow better than others, right? And that are gonna have higher fitness than others taxa. And this is gonna lead to more convergence. If you have two different habitats that are very similar to one another, the selective pressures they will experience are going to be also more similar, right? And that can be a force that will help to generate convergence in community assembly. On the other hand, there's also a wide range of ecological processes that are stochastic in nature, right? And it will lead to more variable community assembly. You have, for instance, dispersal. It's clear that you can only find in one habitat those taxa that were able to colonize the habitat. And the process of dispersal, and this is very true in microbial communities, is gonna be highly stochastic for instance. Think of the bromeliad tanks that I was telling you about before. It is clear that if a specific mosquito arrives and lays an egg into one water tank, but not another, then the bacteria that it carries will colonize that habitat, but not the other, right? So that is going to be a force that will lead to variability. Of course, micro-populations, even though they're very large, they're still made up by individual cells that have to divide and have to die, right? So you have births and deaths, and all those are stochastic processes too. So finally, you have mutations. Evolution will always be a force, and microbes evolve fast. They have relatively large mutation rates and large population sizes. So mutants emerge quite rapidly. And of course, mutations occur randomly in the genome. So that's another force that will generate variation, could generate variation in community assembly. And finally, one that is very interesting is historical contingency, right? One of the very salient features of microbes is that as they grow, they cause dramatic modifications in their environment. For instance, I don't know if anyone here has had experience with growing bacteria in the lab, even for other reasons than doing microbial ecology. But if you have, even for cloning or whatever, you probably took a colony, say, of E. coli and grew it in medium that, for instance, could contain glucose as the main carbon source. You grew it overnight and you harvested the cells the next day. So if you've ever done that exercise, you should know that in the process of just growing from one small colony to a very large number of cells, the environment that is left out after you harvest your cells is completely different from the one at the beginning. In just 24 hours, these bacteria have turned an environment that may have contained glucose as the only carbon source, for instance, into an environment that will contain a large number of other nutrients. And that's not the only way, right? In which microbes can affect their environment. Micros can engage in microbial warfare with one another. They can release antibiotics and colicins and other toxins that can kill other bacteria. They release them to their environment as they grow. They also modify the spatial structure of their habitat by forming biofilms. And they also secrete enzymes to the outside world that can break down complex, the physical structure of the environment with all kinds of complex polymers around them. And the collective modification of the environment that microbes cause as they grow leads to another ecological force, if you will, that is a combination of both stochastic and determinism, which is historical contingency. And we can visualize historical contingency quite easily, right? Imagine that you have now two different taxa arriving in an environment, right? And only this one experiences can grow in that environment. So you have very strong fitness differences between them. But as this taxa grows in the environment, it'll change it, right? And as it changes that environment, then the other taxa now might be able to grow. Maybe even a taxa that could not grow in the original environment could potentially grow after the first one has been growing in it through the environmental modifications that it applies. And that will mean, of course, that if you have two habits that are originally very similar from one another and they get colonized by different taxa, because as we said, that first part is quite random, each one of those taxa will modify the environment in a different way. And now those environments won't be similar anymore. And they will present different selective pressures that will then impact for their colonists that will arrive later, as well as just the overall physical and chemical structure of those habitats. So what's important here is that some of those forces like selection can lead to reproducibility across habitats that are identical, right? Of course, not habitats that are not, right? At least when habitats are very similar, selection is a force that will lead to more similar communities and different conditions, whereas sense and historical contingency can push you away from similarity and may cause more dissimilar communities. And the complications, of course, that all of these psychological forces are acting in nature at the same time. And they're very difficult to disentangle mechanistically in the wild. So one of the questions we are trying to, one of the approaches we're trying to follow in my lab is to see if it is possible to recreate the process of micro-community assembly under conditions where we can study it and we can disentangle and control the levels of variation in the laboratory, right? And our idea would be that if you want to ask reproducible micro-community assembly is, it would be very helpful to be able to study this in a system where we know, for instance, if our habitats are at least initially all identical from one another or not. This is very difficult to do in the wild, right? Even in the case of the bromeliad plants that I said before, you don't really know that all those plants are really identical. You know they're similar, right? But you don't really know what the selective pressures are in those habitats and you don't know how different they are from one another really, right? And moreover, we wanted to see if it would be possible to have these habitats being as controlled as possible so that we could know what the selective pressures are or at least have a very good idea of what they might be. For instance, if we control the number and identity and concentration of all nutrients, if we can control the temperature in the pH and other physical parameters, such as the geometry of the vessel with the bacteria grow, so we can control all of those things and make them equal across habitats. Then we would be able to ask this question of how reproducible community assembly is in a much more controlled manner. And in addition to that, what we seek is a system, an experimental system where we can also control a lot of the ecological processes that are very difficult to know even in natural habitats. For instance, in the case of the bromeliads again, we don't have access to entire colonization history of those habitats and we do not know if some habitats were colonized by different set of bacteria than others, or this is very difficult to really make that inference. But in the laboratory conditions, we can inoculate migrants at no time intervals. We can also know how many cells are arriving and with what's frequency, from which regional pulse of species. So all of those things are things we have under control. We can also control how connected the habitats are if there's migration within them or not. If there are any population, what's the next we can impose them. So we wanted to have, in other words, an experimental system where we can study this question of how reproducible community assembly is under conditions where we can eliminate a lot of the sources of variation that would occur in nature across habitats so that we can really get to the bottom of what are the intrinsic processes that lead to variation at the community assembly level in an experimental system. So if we had a system like that, the question is if we could predict a community assembly. So to address this question and give you a sense of what is the experimental pipeline in our lab, we start, we use a technique that microbiologists call the enrichment cultures and we do a high throughput. We start from natural samples. For instance, we could take a leaf of a plant and then we could stick that leaf into a test tube. And at this stage, we are treating our samples with antifungals and other drugs that will eliminate the growth or inhibit the growth of eukaryotes. For no other reason that we wanted to, these are the very first experiments we did. So we wanted to make them as simple as possible and be left only with the bacteria. And with that, we can take all of the bacteria that live here and we can filter all of these plant particles and we are left with what's essentially only the bacterial component of the community that we had on that leaf. And now what we can do is we could take that large initial pull of species and sample from it and put it into a bioreactor which is a very small, you can think of it as being a very small test tube. And we can add some migrants into bioreactor that contains a synthetic medium. This is a growth medium that we have created in the lab where we know exactly what all of the nutrients are and we know exactly the amounts that we're providing. And then this is for those of you who know more, this is in this experience I wanna tell you about, this is M9 minimal media. And I'm gonna start by using glucose as the only carbon source for doing in the future. You'll see all the other nutrients as well. But say that you have a minimal media where this is limited by carbon and it has a single source of carbon here. So this is the simplest environment you can think of. This is a liquid environment. There's no special structure to begin with and we are at the single growth limiting factor which is in this case, glucose imagine. And now once we colonize that habitat from the community, we let the cells grow. We typically let them grow for 48 hours. And after 48 hours, what we do is we take this, so some cells will grow, some others will not. But then those that grow, we will take a small sample from here and we'll inoculate another habitat just identical to the one we had on the first day. So we again, we have another little test tube when we put fresh concentration of all the nutrients the glucose and all the mineral salts everything else we're adding. And then we let that grow again. And then after 24 hours, 48 hours we'll take a small aliquot and put it into a new test tube where we were playing show nutrients and we can keep doing it again and again. This is called serial passaging and it's a mechanism. You can think of it as a seasonal environment with seasonal bottlenecks as well as addition of new nutrients. And then at the end of every 48 hour period we use 16S commuter level sequencing to take a sense of what cells are present what species are present and what their abundance is. So again, we do this at the end of every 48 hour period right on the ground community before we apply them on to it. So I guess the first question you could ask is if you do this simple experiment and again, you're looking at the assembly of a community in an environment that contains a single growth limiting research you might not think that more than one species might survive. The first question is will you get a community if you do this? Or are you gonna get a single species that will out complete everybody else? Well, let me give you an example. The answer is no, we always get communities even though we only have a single growth limiting research but the way that the dynamics look like, right? On day zero, we have a very large community and I will show you some examples of what this community look like in a minute. It's again, each color represents a different genus and the width of that of its band represents the abundance of that genus in the community. All this gray stuff you see over here these are species that are so low abundance that we cannot really show you here all of the bars that you would need to in order to see them. And now what we see is that after 48 hours community has changed as you would expect because some of these taxa could grow but the other ones could not. And then as a function of time every single time we do one of these 48 hour growth periods community composition starts changing and after about eight to 12 transfers you see that community composition has reached the state where every day after every 48 hour period the communities look very much like they did the day before. So they reach a state of, we call this a state of equilibrium understanding that of course there might be even slower time scales here we're not capturing but we have actually done experiments now where we have propagated communities for longer time but for the rest of this talk and the next two let's just assume that after once the community is stabilized here that this will be a state of equilibrium and as I will show you we have probably the last lecture we have solid evidence to believe that this is actually very close to an actual dynamic equilibrium. All right, so this is the system we have let's go back to the question we're asking like how reproducible is my community assembly in these habitats that are so well controlled? And I think here the idea is okay so a system is gonna be reproducible if every time we do an experiment we're gonna get the same result and if every time we did the same experiment we've got different results then we would not be able to predict media assembly, right? So this is what we did, right? We take from the same species pool we inoculate eight replicate populations a replicate habitats that are identical to one another and we are going to propagate them in the exact same manner as I showed you before and I'm gonna show you the community assembly we observe after about 80 generations or 12 of these transfers. Again, so this is the outcome, right? So if you look at community composition of the species level you find that even though these eight habitats are identical to one another as identical as you can make them, right? Even though they're all colonized from the same regional pool at the same time, right? So even though we're eliminated all sorts of variation to the maximum extent we could, right? I mean, of course, you still have to factor in human error and these are experiments like nothing is perfect but this is gonna be much more reproducible for instance, the bromelia tanks I showed you before, right? This is well-controlled laboratory study and as you can see here, each of these times you did the experiment you basically got a different outcome, right? Here we're showing species, in fact, genus level, it doesn't matter and again, different colors represent different ESVs and you could see here, I don't know if you can resolve this here but these are two different shades of red, right? Or three different shades of red. So anyway, you can see that communities are not the same, right? Even though you did this experiment multiple times. However, we also noticed that if instead of grouping all the reads that we were sequencing by species we grouped them by the taxonomic family that these belong to. What we find is that when we do that then community assembly becomes much more reproducible and that all of these eight replicates contain ratios of the same two dominant families. In blue, this is center of active Asia, this is the family to which the famous bacteria E. coli belongs to and in red here, this is pseudomonade Asia which is another very common in this case, a very common environmental bacteria you find them in the soils and plants and also in the human body it could be some members of the family could be also a pathogen. All right, so you see that when you do the experiment eight times, the results that you observe are very reproducible at the family level but not reproducible at all at the species level. Okay, so this is if you use the same inoculant, right? But what if you use different species pools inoculating its habitats, right? How reproducible would things be then? I mean, I think there's all kinds of reasons why you would imagine that this should lead to less reproducibility, right? Because its habitat now is being colonized by different group of bacteria. So if anything, you would expect things to be less reproducible than the way before where all these habitats were colonized by the same bacteria, by the same inoculant. So we went and collected samples, environmental samples from soil plants, aquatic communities, like soccer field and various other soil plant and aquatic environments around 20 miles radius of our lab. And if you look at, this is at the, if you can see this here, the different communities were quite different from each other, right? I'm talking about the soil inocula. Even when you look at this at the other level, you find that there are, you know, they contain very different species and abundances and taxonomy groups. Okay, hold on. All right, so now when we repeat the experiment that we did before, but now using these 12 regional species pools that again are very different from one another and we process them through the same pipeline, we inoculate them into a different test tube and all of them contain glucose and zinc carbon source that then grow on and you incubate them for 80 generations, just as we did before. What you find is that you have that each of these inocula lead to different communities, right? And at the species level, which is, you know, not surprising because we are using different pools of species, right? Yet when you do the same thing we did again and look at assembly at the family level, now you find that communities assembly are still very reproducible even though the inocula were used in each one of these wells is different from each other. So this is painting this picture that community assembly is actually very reproducible right at the family level if conditions among wells are the same, right? Even when you're putting different sense of taxa on each one of those wells. And this is another way to visualize this data, right? At the start, each of these communities had very different fractions of these two dominant families in Terraciraeusia, Sulmona de Asia, as well as other families. But at the end, they have converged to a very similar location on the simplex highlighting the real reproducibility that one sees in our experiments, right? So all of this that told you about prompts three questions which are going to be the subject of the three lectures I'm giving. The first, just going to be the rest of this talk I'm going to be talking about this first point why so many species coexist in a single-limiting resource. Tomorrow I will be talking about why community assembly is so convergent to the family level, what does that mean? That high levels of taxonomic community assembly is more reproducible. And finally, the last day, I will be telling you more about why is community assembly so variable at the species and genus level, right? And this is going to try to give you explanations for these three kind of interesting results we've revealed, right? First, many taxa can actually coexist when used to play a single-limiting resource. And second, that the composition of those communities that will form are going to be very similar at the family level and very variable at the species level, right? So today I just wanted to devote the rest of this lecture to this first question which is why do so many species coexist on a single growth-limiting resource? So kind of the reason why these might be so much surprising is the competitive exclusion principle, right? We are providing a single-limiting resource in the case that occupies in the data I showed you before, it's glucose. And this is a carbon-limited environment, right? Glucose is a source of carbon. So you might imagine that, or at least in fact, that's what I did when I first did these experiments back in the day that I was gonna see just a single toxin that would have compete everyone else. And if, you know, very simple but yet profound ecological theory tells you that's more or less what you should expect, at least that coexistence should be difficult and the conditions where you have a single supply noted there are, of course, ways around it but at least kind of your first knowledge expectation would be that coexistence is gonna be hard. And in particular, that you really shouldn't be expecting to have more species than there are limiting resources, right? So here we're having one. So the solution really, or the idea that we had was that these might be caused by a phenomenon that is very widespread in microbial communities which is methaolic host feeding. And I gave you some examples before about how when microorganisms grow in an environment they can fundamentally alter that environment, right? And perhaps the best way to illustrate how powerful this can be is to tell you a little bit about this one experiment that is one of my favorite experiments of all times. It's a beautiful result. So this is an analysis of a previous experiment that was done in the lab of Julian Adams where they had evolved a single clonal population of the bacterium E. coli in a chemostat. So for those of you who are not familiar with what a chemostat is, so you think it's basically a continuous culture device. You have a vessel where it's a growth chamber where you would have for instance, population of E. coli growing. And then you are constantly feeding nutrients from a medium reservoir. And on the other hand, you are, what's wrong with my mouse? I don't know why. And okay. And here you are also taking out media, right? As well as cells. So you are keeping, for instance, the volume constant of your vessel. So now the faster you flow, the faster you take out matter too. So you are in a chemostat, you can control the feeding rate, which is the same as the flow rate through this system. And you can also control the content of the medium that you're supplying and so on and so forth. So in this case, there was a chemostat that contained a single clonal population, originally clonal population of E. coli. And it was adapting, it was growing this environment for over 700 generations. And as E. coli grew in this environment, again, this is an isogenic population. What it does is that as it optics glucose, it will metabolize the glucose and E. coli has a form of glucose metabolism that is a respiro fermentative, right? So to grow fast on E. coli, it needs to grow fast on glucose. E. coli needs to partially ferment it, right? And what it will do is that it will metabolize the glucose and produce part of that goes through the entire TCA cycle but some part is just driving into overflow metabolism. And with that results into the production of relative quantities of fermentation by products like acetate, glycerol, succinate, pyruvate and other organic acids that are produced as E. coli grows on glucose. So while it grows on glucose, it releases other nutrients into the environment. And as it does that, it changes the biochemical and nutritional composition of that environment. So as E. coli grows in this glucose in the chemostat, it's transforming an environment that where it has a single growth limiting resource into an environment that contains other niches as well. So that leads to in their study to in a relatively short time about 750 generations, the originally isogenic population of E. coli had diversified into an ecosystem. And so on the one hand, they observed that E. coli had that there was a strain of E. coli that had evolved a stronger growth rate in glucose and a larger optic rate of glucose. But an avoidable consequence of growing faster in glucose is that you're gonna have to be more wasteful, right? And then you're gonna have to create more acetate and more glycerol and release it to the environment. And that enhanced concentration of acetate and glycerol that is where the cells are basically stewing on led to the creation of niches that were occupied by other derived strains that had adapted to consuming acetate and glycerol. So in other words, in less than 1,000 generations, E. coli had diversified from a single clonal population into a small ecosystem consisting of three crossfeeding strains that coexist with each other. And again, this is the power of crossfeeding, right? You can give them bacteria single growth limiting research but they're going to change the environment and create other resources that provide the substrate for other cells to grow, potentially leading to more diversity that you would expect naively. So to get a sense of what a crossfeeding could be a ecological force of importance in our habitats, we did experiments where we took, in this case, one of our communities and we isolated some of the dominant members. And what I usually hear each of these circles here are a colony of a different species. And you could see that here that you have different morphologies of these colonies and that is indicative of different species present in the community. Each species contains its own, when it forms a colony, the colony has a different shape, right? So we're able to separate and to isolate the members of, for instance, this community here which contained three coexisting taxis of the family interactoricae, enterobacter, rautelans interobacter, as well as one taxon of the genus pseudomonas. And one, which is the stenotrophomonas which we were not able to isolate because it doesn't grow on its own. It requires the presence of its ecological partners in order to grow in this habitat. Okay, so first of all, we find that all of these species can grow. There's four that we were able to isolate are able to grow on glucose as the only carbon source. Now, the question we asked then is can they also grow on the metabolic secretions of the metabolites released by the other species? So we did a very simple experiment. We took each of the isolates and grew them on glucose for 48 hours. And after those 48 hours, we took the medium that was left over and used it as the substrate on which the other taxa were put. So we would take another species and see if it could grow on an environment that contained glucose no more. There was no glucose left. It was completely consumed, but contained all of the metabolic secretions by the former, right? And we grew all of the taxa on the metabolic secretions of each other. And the result was quite what striking, right? Here I'm showing one example. This is Citrovacter, one of the four members of this community growing in the secretions of enteroactives. This is another of the members. And on gray, here I show you the growth of Citrovacter in glucose. You see that it has this very sudden rapid growth, which is caused by the overflow metabolism. When Citrovacter grows, it uses this fermentative pathway. And during that time, it grows really, really fast. And then it reaches this first plateau, which it reaches when glucose runs out. And then it starts, I'm talking now about the gray curve here, Citrovacter will start growing on its own secretions. But if you grow it entirely on the secretions of enteroacta, you will see that the growth rate will never be as large. I mean, the slope of this growth curve, which is here in black, will never be as large as the one in glucose, but it can still grow quite robustly and it can reach high yields, right? What this shows is that there's still, even after glucose is consumed, and very large growth potential in this bacteria on the byproducts, right? There's that crossfeeding can be quite substantial in our communities. And we're repeating this experiment for every possible parent in this four. And we noticed that all four of these stacks that could grow on, all of them could grow on glucose, but they could all grow on each other's metabolic secretions. And I'm gonna tell you more about this in the next lecture tomorrow. But before I move on to the next, I just wanted to highlight how important crossfeeding can be even at the level of the whole community. Here we are plotting the biomass of the community as a function of time. And I am plotting it as well as the amount of glucose in the environment. So we measure the amounts of glucose present at time zero, which is, there was a 0.2% glucose in the environment. And then after 24 hours, when we took the next measurement, we quantified the growth, the total amount of growth of 95 different communities that had spontaneously assembled in glucose environments, whereas there's other 95 dependent experiments from different inocular. I will notice that of course, as you would expect, there's a lot of cells, right? Cells have grown on glucose. But after 24 hours, glucose was completely gone. It has been exhausted. Yet growth continued strong for another 24 hours. And in fact, now we know that growth had, that glucose has been exhausted even before those 24 hours, depending on the community, typically takes between 12 and 18 hours for glucose to be exhausted. So the total amount of growth on the byproducts, now we're quantifying this to be about the same as the total amount of growth that that are causing glucose, right? Total amount of biomass. And what we can do then is, when we did this in collaboration with my colleague, Bankat Mekta, and what we were trying to do is, see if we could incorporate this cross feeding into the MacArthur consumer resource models. And the ones that will tell you that you should only expect to see a single species in a single growth limiting resource. And indeed the simulations recapitulate our findings. So the only difference really that we're adding is this term here, which incorporates the metabolite production through cross feeding. And the rest is exactly the same as the MacArthur model. And again, we observe that when we set the cross feeding to zero in an environment where we supply a single growth limiting resource, there's a single species that can survive. The other ones are competitively excluded. But when we assume that the cross feed, just by using random matrices that capture the amount of cross feeding, we find that there's generically multiple species coexisting. And the number of tax you see is about half, typically on average, the total number of resources that can be supplied through cross feeding. All right, so I've given you some evidence that cross feeding is very important, that the environment is no longer a one resource community after even a few hours of growth. And that is very likely giving us the multiple species being able to coexist. But there are other potential factors that could lead to coexistence even without cross feeding. And one of them is spatial structure. You might find that different taxa are occupying different spatial niches within our habitats. So all the experiments that I've told you about were done in stable liquid environments. So it's liquid environments that are still, we're not shaking them or anything like that. And when that is the case, then you may imagine that different bacteria might be occupying different strata on either some taxa could be, for instance, living in air water interface, others could be living in the solid liquid interface on the bottom and so on. So we wanted to get a sense of to what degree this would be important. So what we did is that we repeated the experiment by shaking our growth vessels vigorously, therefore removing as much of spatial structure as we could. And I just have to say also, we had no evidence of biofilm formation, at least suggesting that that also was not an issue. And every single time we've done it, we've observed coexistence of multiple taxa, right? So even when you remove spatial structure to the best of your ability, then multiple taxa can still coexist together on a single supply resource. Another potential possibility is temporal niches and that could also lead to coexistence. And to understand why, let's go back to this community that we have been talking about that we have kind of dissected before that contains this four taxon, citrobacter, citrobacter, enterobacter, surmonas, and robotella. You know, one possibility could be that even if cross feeding was not playing an important role, as taxa grow on glucose, it might be that different taxa have a growth advantage over others at different concentrations of glucose. For instance, it could be that citrobacter had an advantage at very high glucose concentration, but then as the media, as the community starts depleting the amount of glucose, it will drop below the level where it has an advantage. And it might be that later on, an enterobacter in this period here, at this level of glucose concentration, enterobacter has an edge over the others, a competitive advantage, or it could be that then later surmonas and later robotella, right? So simply by depleting the amount of glucose as the cells grow, that's another way in which bacteria are modified in the environment, and that could potentially to go exist. As in fact, in some cases in yeast, it has been observed that that is a mechanism that is important. There's some yeast that are better at growing at high glucose concentrations, others are better at growing at lower glucose concentrations, right? So that could lead potentially to a stabilization mechanism. So we wanted to test this idea. So what we did is that we took those four taxa and then we grew them at different glucose concentrations and measure the growth rates at those concentrations. And what we did then is we parameterized the growth rate as a function of glucose for each one of these four taxa and using a monot model. And then we just created a very basic consumer resource model that contains this parameterization and simulated the growth of our communities, again, assuming this periodic supply of glucose every 48 hours followed by a dilution. So we parameterized this and asked whether in under the condition of our experiments, you would expect this four taxa to coexist and there's no, right? You can even hint, just qualitatively looking at the data that there really isn't a very serious crossover, right? Citrobacter tends to grow on glucose better than the other system, the other one that crosses over, but in terrobacter, so the monosum roll-tela, their growth curves are very parallel to one another, right? So it's not like they're occupying different concentration niches that with the depletion of glucose over time might lead to coexistence. Another possibility is acidification. Again, it's another way in which bacteria can change the environment without necessarily invoking any cross-feeding. So as I was telling you before, the production of acetate and other organic acids and lactic acid in particular will bring the pH down, right? And if you look at the various members of this community in terrobacter or l-tela-citrobacter, these are in terrobacteria, which are fermentative and they all produce different organic acids. And when they do, the pH of the environment comes down, right? So one possibility was that as these cells are growing, the pH is dropping, right? And if the pH is dropping, it might be that some tax are better than others at different pH, right? So this is another mechanism of niche construction in this case that is not mediated by cross-feeding, simply is lowering the increase in the acidity of the environment slowly. And it could be the different tax that have a growth advantage of different pH values. However, what we find is that even though each one of these taxine isolation will lower the pH of the community of the habitat, the community itself does not, right? So that when you look at the pH over time for a community and here are some examples, the pH rarely drops below six and in most cases even it remains very, very similar over time to what it was in the beginning. Again, the individual species themselves do acidify the environment, but the community itself does not, at least to any appreciable degree. And at any rate, we also measure the growth rates of these taxa at pH ranging from six to seven, which is where the pH lies and over the 48 hour duration time and there's really no crossover point either for that. And finally, what we've been also looking into is cell death, right? So one possible mechanism that is not cross feeding but could also lead to coexistence is cell death. And there's two different ways in which you could see this. For instance, you can think of, well, we're growing these communities under serial growth conditions. So we are growing them for 48 hours. And in that time, cells have time to grow, but also they would have time to die, right? So if cells are coexisting with each other, it's gotta be true that they have the same fitness, right? So basically every species that is coexisting in stably in these environments will have to double exactly 6.7 times every 48 hours if they want to coexist with each other. And it is possible that that is done because some cells actually grow better in glucose than others, but then they die, right? And then the two cuts up at the end, right? So that's one possibility. But another possibility also has to do with environmental modifications. If cells die, they will also spew out their contents out into the environment, right? So that can also create niches, but I wouldn't call that cross feeding, right? It's not really cross feeding when you're dying. So at any rate, we wanted to explore this question. And we asked how much death we observed in our habitats as a function of time. And there are various experimental techniques you could use to monitor cell death in culture. There are both dyes that will stain to the polarized membranes that, and I don't remember the prizes you see an unviable cell. And also when cells lies, they leave these kind of corpses behind that you can also detect by face contrast microscopy. So we looked at micrograph after micrograph of these communities to try to quantify the number of cells that were either the polarized or dead or licensed after a different time points. And we observed this for two different communities that the fraction of that cells is relatively small and we don't really see, in one case, we saw a little bit of a change over time, but in the other we really didn't. So it doesn't seem like cell death is gonna be a major contributor to coexistence in our communities. We're not seeing anything massive, at least. All right, so in this first lecture, I introduce you to the problem that we're gonna be talking about for the next two. And I've been also focusing on, I've told you these three questions that we wanted to address. And I've been focusing on this one, right? And why so many species coexist on a single limiting resource. The answer that we are finding is that by and large, the main reason is metabolic crossfeeding. There's many other contributors and I'm not claiming that acidification or that other factors could not contribute as well. In fact, for example, if cells die, they will release their content, even though they're not many cells. And that is part of the environment, right? So basically everything these cells are doing affecting the environment will affect their coexistence. But crossfeeding is a very, very large factor, right? Like about half of the total biomass that we observed in our communities originates from molecules that have been released to the environment, for nutrients that have been released to the environment by the bacteria that grew on the supplied resource, right? So it's from what we have been able to tell the dominant factor in our communities. So the summary of this first lecture is simple, right? First, we have observed that a very large number of taxa can coexist in CLE passage environments with a single supply limiting resource. We find metabolic crossfeeding being a primary responsible for that coexistence. And we also find that communities that are assembled in simple habitats can exhibit reproducible assembly at high levels of taxonomic organization, a family or higher, but they're very variable at lower levels of taxonomy. And this behavior is reminiscent of this metabolic convergence versus, or functional convergence despite taxonomic divergence that people have reported in nature. But we have a better chance of understanding this mechanistically because these communities have been assembled in habitats that we understand. And this is something that I will explain tomorrow and on Wednesday and we'll get to it in more depth. So I just wanted to close by thinking everybody in my lab and who are the people who have been doing this work. And yeah, that's all I have to say. If you have any questions, I'm happy to take them. Great, thanks a lot, Alvaro, for the fantastic lecture. So there are a few questions from the chat that I can start reading from you. And in the meanwhile, if anyone wants to ask a question, please use the raise and the tool of Zoom. So there is Liao who is asking about the crossfeeding whether you have measured metabolomics in the spent medium to determine which metabolites are key mediators and that it might be comments that might be interesting to compare the number of species in the community and the number of such key cross feeding mediators. Yeah, that's a really good question. We have done metabolomics for some of these communities. And so the dominant byproducts that we observe are acetate, succinate, pyruvate, and lactate. And then we see a large number of other byproducts relatively low abundances. And this is doing LC-MS, mass spectrometry. And there are, of course, a large number of resources right in the mass that are detected through mass spec. But as I said, the key here is not just the number of resources you have, but also the abundance, because many are relatively rare. Now, in simulations we've done, it is true that even resources that are relatively rare can be critical to stabilize communities. So in principle, you could have more, so you could have just a handful of resources that have abundance and a large number of resources that are low abundance. And those low abundance resources, even though they might seem unimportant, can actually be quite critical, at least in simulations for coexistence of fairly decent number of species in our habitats. So I completely, I think it is a very interesting question, is how many resources do you need in order to have coexistence? And we've done also experiments in our lab where we went from just having one resource to two or three. And what we find is that the richness of our communities doesn't grow, it barely does, right? So even though you're doubling or tripling the number of resources, the number of taxa you see barely grows at all. It does a little bit, but only in a statistical sense, and it's in most cases, there's no increase in richness. So there's, I suspect that there's other factors that limit diversity in our communities. And we still try to, we have some ideas of why this is, but we don't have any actual evidence at the moment. Thank you, thank you. That's fantastic answer. Great, so there is one question from Lorenzo. Hi, I wanted to ask you a bit more about the eight communities developed from the same sample on the same resources. And I wanted to ask about if you have any data on temporal stability, so the stability of the composition in time instead of on different samples? Right, so yeah, we do. I hear I'm showing one, I mean, I have other plots I could show you, but this is quite general, right? So this is, I have, and maybe I can bring a slide for next time and show it at the beginning. But yeah, no, we've done it. And in fact, it's in the paper if you go to the supplementary material where we show data like this for 12 other inocula, that we find is that there's this initial, actually it's quite remarkable. We tend to see an initial increase in the abundance of interactive Isia, but then as a function of time, it stabilizes and after about 12 transfers remains constant. And we have another paper, we've done this for 18 transfers instead of 12. And yes, once they reach about, it depends on the composition, but between seven and 12 transfers, communities will stabilize and remain constant for the remaining of the experiment. And we finally have an experiment that we just did it, we propagate 12 communities for a year, and but we're still processing the data. So yes, we have data on it. And it's in the supplement of this paper, which I don't know why, sorry, this is from an old slide, it's 2018 in science. And you can see there in the supplement, you can see the results that look exactly like this, but for other communities and they look very similar. Thank you very much. Great, there is a question from Martina. Hi, I have a question regarding the spatial structure. And I agree that you can have coexistence without spatial structure, but since you have different results when you shake or you don't shake, do you think that the stratification might be important, especially because you have some, I don't know, byproducts of species that maybe are doing an aerobic metabolism, that maybe are at the bottom and... Right, so we, that's a good question. And here, this is, they're color differently, but this is, so I'm sorry for that, but this is, this is in terabacteria and this is to the monads. So that at the family level, the results are not that different, right? They're very similar whether you shake or not. It's just the coloring is different, but sorry about that. But we find that actually they're flipped. Here, blue is red, red should be blue. But yeah, no, it's, we see the same. This is in terabacteria, that's what I'm saying. But we have now done experiments in deeper wells where we're a different geometry. And when you do it then, then you do find that the respires have a hard time. And I think it's because of oxygen depletion that cannot be supplied fast enough, right? So I mean, clearly if oxygen cannot get through, then the respire is gonna have a hard time because they don't have like receptors and fermentative metabolism is going to be favored. But at least in these experiments that I'm showing here, like when you do this in 24-watt plates or in not very deep 96-watt plates, then the results are very consistent and doesn't matter whether you shake or not. The outcome is very similar. But your point is well taken. And I think if you have a deeper well where there can be an oxygen gradient, then by all means, I think that spatial structure can be quite important. Thank you. Maybe you can ask me just a quick question. The metabolomics were targeted or untargeted? So we've done targeted and we've also done more recently untargeted too. Okay, thank you. Great. The next question is from Kiseyok. Oh, hello. Hi. I just put the questions in the chat, but do you have the comparison between the effect of serial passaging? Because if you do a lot of serial passaging, then it is putting a selection pressure on the microbes that prefer the carbon, the single carbon source. Sorry, I don't think I could understand the question. Can you repeat, please? Okay, because in your experiments, you do like a lot of serial passaging, more than 80 generations. So what would the effect of those serial passaging be on the final constitution of these microbial communities? Because I think those procedures would impose selection on the microbes. Right, right. I was curious, if you didn't do the serial passaging, what the reason... Even if we had done a single batch and never... Right, no, that's a really interesting question, right? Because one alternative and with something we're exploring now would be to have just done a single batch, right? And then let cells basically stew in there, right? And die and whatever, right? There's that is something we're exploring now because I think by... In fact, we know that experiments would do serial passaging every... So you could think of that experiment as a limit, right? When T tends to infinity, right? Of serial passaging, right? Of an incubation time T, right? And now I can tell you that if you do the... We've done experiments where we incubated only for 24 hours instead of 48 and the pseudomonas are gone, right? We only see intervectivitis and diversity plummets, right? If you go from 24 to 48, then you see pseudomonas appearing, right? Because the pseudomonas are the primary consumers. This is kind of a spoiling tomorrow's lecture, right? It's okay. Pseudomonas are the primary consumers of the byproducts in these communities and they grow primarily on the second half of the 48 hour, right? So I suspect that if you leave it longer and longer, right? Then you're gonna start catching up other species that may grow more slowly, right? But that may be... Maybe they won't die, right? And they will... I would expect to start seeing cell death if you let the cells there for very long periods of time. And they will have a succession of bacteria. Now the complication though is that bacteria can start mutating very fast. They can enter this gas state and start mutating really fast when they start, right? And there's other issues that could occur when cells are starving from profit induction to... I mean, evolution will become a more important role, play a more important role, I think, in our communities. So yeah, it's a really interesting question. It's one that we want to explore in the future. And we've started by looking at just going in the other direction first, looking at what happens if we cut short incubation time. But my expectation is that as you increase the incubation time, diversity is going to go up, right? And you're gonna have more access together. But it's just a hypothesis. We haven't done the experiment yet. I have another question. So if like when you are doing 16S RNA sequencing, is there a way to rule out the dead cells in the tube from being detected as live cells? I'm having some trouble with my mouse. Okay. Can you repeat the question? Yeah. Yeah, when you are doing the Amplicon sequencing, would there be a message to rule out the dead cells as being live cells? Right. That would be difficult just with 16S. We, for the experiments that I've shown you today, we did do this, right? We measure, we did microscopy to get an estimate of how much cell death there was and how much that was contributing to our community assembly. And we found that it should have a very small contribution, right? There's very few cells that are either dead, dead or in a metabolically-arrested state that likely would be dead, like if you're using live dead staining. So yes, but I agree that if you did this, I think the only thing I could think of is this. This will be just a microscopy in conjunction with 16S to get a sense of how important cell death is. At least cell viability, maybe more than death. Death is always difficult to harness than it seems, but at least viability or use in CFUs might also be another way to do it. Okay, thank you so much. You're welcome. Great, there are actually more questions, but we are sort of out of time, but I think that there will be time for them in either tomorrow or on Wednesday or on the next lectures. So I ask everybody who has questions to keep them and I'm sure that since the lectures are very similar topics, they will be asked in the next lecture. So thank you very much, Alvaro, for this very nice overview. And we'll take a break of...