 And time – we can wait one minute. We are late as per the tradition of the country. I think we can now start. The first speaker of the last day, of the first week is Martina bee Valo, who will talk about the metabolic lens part 3. Does it work? Yes! Alright, good morning again! Jaj prijevam, da je prijevamo všeč, da je zelo izgleda, počakaj, na tem, da je od všem všem, pa je dobro vse, da je dobro vseč, na tačno, da je dobro vseč, na to, da je zelo izgleda, na to, da je dobro vseč, na to, da je dobro vseč, na to, da je dobro vseč, might be useful to understand, actually, to address this question that is, how do nutrients shape the diversity and structure of microbial communities? So, but when we think about nutrients, well, it's a bit different compared to temperature or salinity because there is only one, let's say, one axis of variation, temperature and salinity either decrease or increase, okay, salinity maybe you can take into account different souls, but in the way we think about salinity is mainly increasing or decreasing. While for nutrients, while we can think about many ways nutrients can vary and all these ways nutrients can vary actually can affect how species interact and how community assemble. And what are these axes of variation? I mean, it's pretty straightforward. One important axis of variation is the concentration of nutrients. So, how much of each nutrient is available or how much is the total of all the nutrients available. Then another important axis is how many different nutrients microbes can access. So, the number of different type of nutrients. So, whether you have one sugar or two sugars or three sugars. And finally, the last important axis of variation is which resources are available, which type of resources. Let's say here I'm mostly focusing on sources of carbon but the same whole story applies in principle to sources of nitrogen and phosphorus, micronutrients, you name them. But basically there are different ways of thinking about different type of resources. You can think about sugars and organic acids the way Otto was describing them, but also you can have recalcitrant or labile carbon sources depending how easy it is to break them down and consume them. So, and today I will try to go through in a kind of systematic way how you can approach these different axes of variation of nutrients. And the first thing that it's pretty easy to, well, to approach is, okay, we have one resource at one concentration and then, but you have many, so it's one resource available in the environment at a fixed concentration, but there are many types. And we already seen that despite the general belief is that if there is one resource you should have only one species coexist surviving at equilibrium because of the principle of competitive exclusion. In reality you have many, many species coexisting even when you provide only one source of carbon. And the other important thing is that, well, it follows that if you don't have just one species surviving you can have many and this diversity or richness or number of species can vary depending on a carbon source. So, why should we be, let's say, a bit surprised based on experiments that when we provide one carbon source we see many species? Well, recently Alvaro Sanchez lab put out a paper showing that, well, not very surprisingly because if you do otherwise experiments it happens pretty often that if you assemble communities of two species at different fractions in minimal media with glucose at one concentration and you ask what's the most, let's say probable outcome of the outcome is, well, it's competitive exclusion. So, they tried, I don't remember how many of these, but it's not important, but they tried many, many of these per wise competitions and the outcome is, well, you have more than half that end up in a competitive exclusion outcome. So, people were really surprised, we already seen this paper when they discovered that actually when you provide one single source of carbon, again glucose at one concentration, but you start from natural communities. Josh and Nancy here started with communities coming from leaves and soil. They put them in minimal media. They do these daily dilution cycles. Well, at the end of these eight dilution cycles they find that, well, there are many species coexisting. So, here in this plot you have each of these bars actually an inoculum coming from a different leaf, a different grain of soil and you can see that they have many, many species and even replicates of these inocula show, well, different species. So, the question is, is it just glucose? Is it, what is special about glucose? What about other resources? Well, I did kind of the same thing. So, I started with a grain of soil and I put this grain of soil. So, what you do is you go outside, you take this grain of soil, you shake it so that you obtain a bacterial suspension then you inoculate this bacterial suspension to minimal media plus resources. And you do, I did the seven daily dilution cycles. The important thing that this is mainly because we had a discussion yesterday, I was shaking the hell out of them. So, they were really shaking a lot. I don't expect any special structure here. And what we see is that, well, it's not just glucose harboring many, many species, but also other 15 resources. And here you can recognize different sugars and then we have different disaccharides. Citer and fumarate are intermediates of the TCA cycle and then there are sugar alcohols. This is one weird amino acid that, well, it's actually hydroxyproline that is common in the soil and we have also cellulose and starch. So, here the observation is, well, there is a lot of coexistence in any of these carbon sources and you can see that on average there are about 20 species across all these carbon sources. But I think that the other important observation is that diversity varies among all these resources and we have these, for example, intermediate of the TCA cycle harboring about 10,000 species, but then you can go to, like, sorbet or cellulose and they harbore many, many more species, about, like, 30, 40 species. So, in the first, let's say, question that we had in mani is, okay, how do we explain that it's not just that you see many species? Okay, the question is, what's the definition of species? So, here the definition of species is amplicon sequence variant. So, when we, at the end of the of the daily dilution cycles we extract the DNA and we do 16S amplicon sequencing and here each of the, let's say, unit that we count is a sequence that is different from another one by just one nucleotide. And this is the definition. This variability doesn't change if you aglomerate at the genus level, at the family level, or if you change the metric with which you look at diversity. This is just richness, but you can do it with Shannon or Shannon entropy or the inverse of Simpson that basically it started taking to account the abundance of these species. I don't have these plots here, but I can dig them somewhere. What was the second one? Oh, well, okay. Of course it depends the number of reads that you have and the deeper you sequence you can always, in principle, dig up more species. But maybe I can answer a slightly different question that is if you have different number of reads per sample you can down sample at the same minimal sampling reads and you don't obtain a different result. So usually you have problems with different, let's say, number of reads if you sample from very different environments. In this case, since we run all these samples in the same lane with the same machine the same preparation usually don't see big changes and across all these experiments that I've done in which I do things in the lab, I bring them to the facility and then I get the 16 years back the number of reads doesn't really affect how many species you are detecting. But this is the lower bound. In principle, if you go deeper, you might get more. But this requires different lanes, different machines at the moment. We haven't tried yet. But the thing is that I think it's the opposite because in the lab since we're doing these daily dilutions you are selecting for species that can't survive the mortality which is the dilution rate. Yes. I guess from this plot just two outliers maybe is cellulose or at least one outlier cellulose for polysaccharides you can't really expect the same principles to apply because they are broken down in different lengths of the chains and there is a different number of resources hiding inside a big polysaccharide. If I was analyzing this data I would exclude polysaccharides just because they cannot be imported as a whole unit into the cell. You can... Yes. It doesn't change anything but it's just that if you look at the champion here it's a cellulose and that's not completely fair. No, it's true. The second one is orbital which in principle can be imported. But no, I think it's a fair point. You still see kind of surprising things for example the disaccharides. You might say you can actually maybe import them and you might expect more species but it's not the case. This is not just this experiment but other experiments. So the problem is how do you try to understand this variability in diversity? Yes, this is the question. One thing that we did at the beginning is let's look at the molecular weight with the idea that if you can quantify in some ways the complexity of the carbon sources maybe we get an idea of the diversity. Let's exclude these that have very high molecular weight is actually above. Well, it depends also on which type of cellulose you're using but anyway let's do the exercise that Sergei was proposing but let's look at the diversity of the other stuff. Well, you can get many different let's say different richness with basically the same molecular weight and actually as I was saying you might expect that more complex resources like the disaccharides my harbor more species but this is not the case. So we ruminated a bit on this and then we said okay, well maybe metabolism can help us and this is a map that you've already seen because Otto I think had it on but yeah, this is fairly complex and we didn't want to go into metabolic models because as it turns out or like you need a lot of data as input and so sometimes they give very weird results so we said maybe let's avoid metabolic models, let's do a very simple thought exercise which it is wrong but it is a thought exercise that by the way might be helpful and it is okay. And from let's say these metabolic maps that there are on kegs the number of metabolic intermediates that a generic ensemble of microbes can produce starting from each of the carbon sources that I am providing in the media and this analysis called scope expansion analysis and what you need to perform it is the generic ensemble of microbes and we said oh we are in soils more or less that's the species that we get but it's not very defined then some currency molecules so you need an ATP, you need a a COA so the cell needs to be able to function and then you input the supply resource and in this case is again either glucose or sucrose so on and so forth and so maybe this is a very simplified map that you get and this is not actually the map that we get with this exercise but the idea is that well you can basically place all these carbon sources in a map and try to understand how many metabolic intermediates you get starting from each of them and you already recognize that there are some especially so all the sugars so the glycolytic substrates are all in this upper part and it takes a while for them to go into the central carbon metabolism while for other carbon sources like the TCA intermediates that are already on the backbone of the central carbon metabolism so the result of this total exercise is a predicted number of metabolites that is the intermediates that can be produced and we are making this huge assumption which is not true probably that in principle each of these metabolites can be released in the environment and can be a substrate for the other species again this is not realistic but it is a total exercise but anyway what we get is that the richness of our communities correlates decently with let's say we can say the position of the carbon source in this metabolic map I see that you are very disgusted tell me can't hear you so how do you mean the intermediates that can be produced for citrate if you feed into the TCA cycle then you can also do gluconeogenesis so I don't really understand but you don't produce starch in principle in this kind of toy network how many of the things you can go to I see but you still do gluconeogenesis with citrate ok, I see thank you it's again this is not extremely realistic but it was trying to get an idea whether the position at which the resource enters the metabolic network that basically is also dividing between glycolytic and gluconeogenic substrates can give an idea of the number of species that you can support in the media yes ok, so citrate citrate fumarate together but compare citrate and glycerol I would imagine there is just other than glycerol everything is the same why you have 10 10 metabolize more for glycerol than for citrate I think because it takes a bit to get into the into the but all the intermediates the exact amount citrate, you see it's in the TCS cycle glycerol takes one step to glycerol's free phosphate which because it's part of metabolism you have to make membrane so the thing is when you look at these maps different species so we don't have just one species so what makes a bit more so what makes actually this expansion for the glycolytic is that from different if you get different species the pathways to go from but the glycerol too because you need everything else other than glycerol you have all these substrates anyway for regular metabolism no, no, you need them but let's say if you start from if you start from glycerol and you add acetylkoen ATP that you get all of them so just one but how do you get 10? why only one the difference between citrate and glycerol on the x-axis I think there's like 10 but glycerol also goes through here so honestly so this is something that we need through the maps and it's possible that the reason why glycerol is more is because there are different pathways I'm not sure about glycerol I mean OK sorry? so many of the other ones maybe OK, but this is not the map there's a basic set of metabolites that we'll be making anyway make a threshold in concentration no, here no there's no threshold in concentration but then yes, I mean if you and the sucrose just breaks down to glucose I mean I would just say OK, so that's one away but the wise glucose and say citrate so many numbers away because glucose with the gluconeogenesis stops around here, right? so in principle you have more no, glucose one step go to glucose phosphate which is always made but you're so needed to do things so, OK don't get me wrong this is what the analysis we checked a bit the different pathways if you are it seems reasonable we can check again all the pathways you're following some algorithm that I'm OK, I don't understand OK so the basic set is the same and I think that's the whole idea so, OK, we can I think that in general you can agree that with sucre with sugars in principle you have a larger diversity of pathways to which you go to the but different species have different pathways yeah, but for example if you are an entrobacter some bacteria can do both the EMP and the ED and this counts as two pathways so that's why you expand yeah, we can look at the scope of expansion analysis but this is something that we get out and we checked from MetaScik and it seemed reasonable what we got OK, fine question so, actually come I just wanted to say that this is testable you take the spent medium from two different experiments no, spent media experiments are pain pain, but the people do them no, I did them, but I don't think you get you can't really relate these things but you should at least statistically find different secreted metabolites in two very different OK, what I see for example is that if you feed let's say glucose to a species that can't do like they cannot ferment you get less metabolites and other stuff that they cannot use for example, but this is through species and then getting exactly yeah, we can look at spent media experiments I don't think they tell you exactly what we're showing here but again, this was a thought exercise and I know that I would get this so I was waiting for it but let's say, the idea is and we can and I don't know that if you have sugars and you have different species you have different ways of getting from the sugar to the central carbon metabolism if you get the TCA intermediates there's no way you have diversity in the pathways they all entered it's a pan genome so we take which pathway are you following yes, there are different metabolic networks for different organisms exactly so which ones are you following I'm following a bunch of species that were in our experiments and that are in the soil so when we get them from keg it's not just one genome it's a pan genome you can select for soil species on the keg so when you put a number out there you're referring to particular species you have so many species I'm not referring to so what do you do? where Maltos is the 34 predicted number of metare 34 you're just based on an average coming from a different, you don't know all of them because in principle so you have you have five species let's say you have five ways of producing these metabolites you count all of them you add up all of them I add them I don't add up the five glucoses that they can produce but I add the metabolites that are not the same so that explains part of it so you know enough about your species that each one you have a metabolic network that's the thing, I'm not doing that I'm just saying I don't know exactly what I have about my species but in keg you can say can I get a bunch of soil species that might be representative of the sample that I have and I consider all of them ok, great let's move on from this so the question is ok, what happens if you add more resources to the game and how we did the experiment was ok, no, first I'm telling what you see so here we're going from one to two resources and then from two to many resources we still are at the same concentration of carbon and we still have many types of resources what we see that was quite unexpected is that going from one to two resources we don't see an increase in diversity but actually an averaging effect that it's still bothering us and the other thing that we see is that richness increases only modestly but linearly with the number of supplied resources ok let's go from one to two so what we did in the experiment was starting from these 16 carbon sources and mixing pairs of them like in a tournament and so for example glucose and hydroxyprolin and then we started saying ok, when we mix them what's the expected richness in the in the mixture of glucose and hydroxyprolin and here since we are keeping the sorry, since we are keeping the total carbon concentration constant it means that there is alpha of this carbon that is due to proline and alpha of this carbon so this is the richness that we observed in single in single resources so hydroxyprolin is about is 11 species, glucose was 24 species and we ask ok, if we combine them what's the expected richness so if we believe this pathway thing where we can say that each species so each metabolite then corresponds to one species what we can expect is that when we combine these two resources we get the union of the species and this is something that was actually observed in a previous study that is published on PNAS where they were looking at communities from the phycospheres and they were looking at combining different carbon sources so in the union would say about 30 species well the other possible prediction is that while you don't exactly get the union but still get the maximum if really the concentration of carbon matters and sets the number of species that can be supported well maybe it's the maximum of these two carbon sources so it's going to be 24 species but then we do the experiment and we found that actually what we observed is not the union not the maximum but actually the number of species of glucose and proline and again this might be a special case but actually when we looked at all the different combinations I think we had about 24 combinations this is through across many of them so that on average when you combine two resources you get the average number of species of the two constituent singles which also so this is a lot to take but basically here you have the number of the richness in single resources the richness in combinations and each line goes to the how the indicates the pair and you see that you get many lines that cross in the middle indicating that the richness in the pair is actually the average richness of the constituent singles so which means that basically if we start plotting the richness as a function of the number of supplied carbon sources what we see going from 1 to 2 is that diversity doesn't really increase so it stays basically the same and I didn't tell you before but then we tested also combination of 4, 8, 15 and 16 species and I'm going to plot how diversity changes as we increase the number of resources in the mixture and it's a tournament so again in principle you can go back to all the different constituent resources so you see that basically region increases only weekly but it's pretty striking that it increases linearly with the number of resources and just as a reference this is the line that you might expect from competitive exclusion so basically we are seeing that there is an offset so you get more species coexisting in one resource but then the way you are actually increasing the number of species kind of follow the competitive exclusion principle because you are adding one or two species for each new resource that you add in the media so of course it is a sample and it was completely random so the way we assemble the 4 is random so again it's a tournament so so I don't have a way of telling whether it matters or not so this is just a collection yeah it's a random and the reason the scatter is less for the higher number is that you ran out of combinations well with 16 resources I can only do one so of course this is I don't remember this is 24 what is 16, 8 so and then we I have all the possible combination of all minus one and then I have the 16 and we did basically three tournaments Is this including the polysaccharize? Yes I think so if you return it without excluding the polysaccharize so if you exclude them here you are excluding this blue dot here so this blue is the cellulose and the start is not that high so somewhere around here and then actually maybe let me go back here so this is where cellulose goes when you combine it with something else so actually even when you exclude them since there is this averaging effect it doesn't change that much the whole picture so but the point is it doesn't increase that much and it's not and I wanted to show that this is not just our result so in the lab of Daniel Segrede they did a similar experiment but starting with isolates so not with the natural communities but they were mixing some isolates and they went to actually to 32 resources and of course you got the fairly raw richness because I started with the natural community that had like 700 species but still you see so the black line is the experiment then these are consumer resource models but let's forget about these two the experiment shows that actually going from one to two so from one to two is where actually you get a decrease in richness and then you start increasing but the increase is very modest so it's not I think this is not just us looking at this result but it's something that other labs have seen and so the question is ok why do we see this pretty striking trend of richness increasing linearly with the number of resources and so we start to think maybe we can look at community structure and get some ideas so and my idea here was trying to tell you what was our thought process that went through this data it took us a while and so the idea is that how did we think about it and then actually I'm going to fill in what other people found a couple of years later that actually was kind of consistent with our thought process ok we have already seen that there are different ways in which we can look at community structure we can try to course grain in different ways you can do a course grain based on resource occupancy so basically trying to understand which species you found in every resource or in every combination of resources and you can call it an habitat generalist or if you find other species that instead are more associated with a single resource then you can call them habitat generalist or let's say course grain based on consumption patterns for example you can use the sugar acid preference that Otto showed 2 days ago and finally you can do the mainstream thing that is try to do based on taxonomy and for example you can look at the family level instead of the ASV level that Alvaro showed it was pretty messy in the mainstream way so this is again data from Alvaro's paper that shows that when you look at communities at the ASV level so these single nucleotide differences you might think oh it's a mess but then when you look at the family level actually you see a higher level of let's say consistency across communities and so we ask ok what's the level of consistency going from single resources to 16 resources so and we took let's say a slightly different approach so we trained the model XGBoost 3 regression model not important on our data in single resources and so actually we have all the data set and we trained the model on a percentage of the data and the input was the relative concentration of the resource and the final abundance of each family that we saw in the community and based on this model we end up finding very strong associations between the family and the resource and then what we did was plotting the relative concentrate the relative abundance of some of the families that showed a very strong association as a function of the relative concentration of the resource that was identified as their favorite is it clear so here you see for the pseudomonas which are one of the abundant families especially in the organic acid and actually drugsy probably you see that relative abundance of this family increases very well with the relative concentration so there is a very nice correlation between the relative abundance of this family with let's say relative concentration of this resource and this let's say enterobacteria are kind of the nemesis of the pseudomonas usually the pseudomonas feed on the byproducts of enterobacteria and you can see that in the same resources actually the enterobacteria which cannot consume many of these TCA intermediates or at least are less good at consuming these TCA intermediates decreases with the concentration of these resources by contrast enterobacteria really like silos and so they increase accordingly to the percentage relative concentration of silos because in all the multi resource communities while pseudomonas decreases then there are other families for which we found associations that were pretty strong for example for the Bacilace they are they reach a decent a higher abundance when you have a lot of cellobios and it's the same for this cellbibronace that can degrade cellulose a decent let's say 0.2% abundance 0.2% abundance when cellulose is present so I've never actually shown this data they are very hidden in the paper but I thought that this might be a good place to discuss them because we see associations and actually you don't really need to so this is actually so you can train a model to find these associations and then you can verify whether there is some pattern and I think this is a very beautiful pattern that is in agreement with the fact that there are strong associations between families which can thought to be the functional unit of the taxonomy and the resource that we put in the media and these associations remain when you have mixed resource environments I ask a question so in terms of contribution to diversity so if you look at the diverse I don't know what is my question but if you look at the diversity within cellulose how does it change as a function of the resource so what I'm saying is how much of the diversity that you see for instance the variation of diversity that you see is due to the variation of the relative abundance I don't think we really looked into that so it might be nice to check It's a very interesting plot and what jumped at me are indeed those two right columns where you only reach any non-zero abundance if this is the only resource and again given that cellulose is one of those culprits and cellubios it may be due to the fact that the growth on those resources is extremely slow you need to sort of break them up so if those resources are given in the mixture and you are a specialist in those resources you will be wiped out by other species so it kind of makes sense did you see it for other polysaccharites I believe you had one more if I start maybe it's a little bit easier degraded I have plots of this for many I don't remember for starch but I think the so we were looking I think the model identified identified something for starch too I can check again yes so this is the relative concentration of the resource in the multi-resource environment so for example is 100% only in one condition it can be 50% again in only one condition but then it can be 25% in the four conditions so this is the colors indicate how many resources you had and then where you don't have it you are not convinced ok, I don't understand the plot anymore you don't understand the plot so this is when you have only only adroxyproline 100% this when you have 50% adroxyproline this is when you have two resources then is eight resources ok, no, no, I understand now but then what surprised me is how small is the bar how small is the variation so the bar is the standard deviation is the standard error and actually this huge bar it means that probably even when there is only cellulose not in all replicates they were present actually so that's the bar because for each resource we had three replicates but then this kind of now back the question so we know that bacteria show strong preferences for either sugars or organic acids and so I was discussed and we saw this before I don't have to go through it so basically if you can assign this sugar acid preference score it's positive and goes towards one if the species prefers sugars and it goes towards minus one if you have acids and this depends on let's say the abundance of pathways for sugars and for acids and since this SAP can be predicted from sugar acid pathways abundance in the genomes well, I chatted with Mati and we said well maybe we can look into it in my communities and what we found is that this is the percentage of sugars enrichment in multi resource combinations so you see that it can increase and become 100% this is 100% corresponds to the minus one so 16 minus one that don't have the acids you can see that the sugar acid preference goes from being close to zero to actually starting to be close going towards more sugar preference and I thought this was interesting because in reality when you look at communities yes, you find several so you never even in the sugars I believe you don't find just sugar acid so species that prefer sugars you also get species that prefer acids which is consistent with this idea that if you provide sugars many species can also excrete acids and these acids can be taken up by the species that prefer the acids but then this fed into another thought exercise that was in our paper that is ok we've seen this map very simplified all the problems of the case that we've discussed before but let's say I think that it's ok to say that if you start from different resources you might have metabolites that are always produced by everyone and then you can have instead metabolites that instead can be produced only by a few resources and so we decided to plot and so this is it's very simple to count if in our scope expansion analysis we found the metabolite that was produced by more than 12 resources then we call it common if it was produced by less than 3 resources we call it rare and if we plot the predicted number of metabolites and we let's say color the bars based on whether the metabolites are common or rare we see that in principle every resource produces this of course every resource produces common metabolites only with the sugars basically you get these rare metabolites it is the picture that I was trying to say before that you get diversity in the pathways only when you have sugars and not when you have organic acids and this is our let's say working hypothesis here and this was kind of we were quite happy when we saw actually that if you do the same exercise with species so here you have these are plotting the various resources these are all the species that we have and then you can ask ok how many times this species is present in how many resources this species is present and you can get again some species that are only present in a few resources other species that instead are present in many many resources basically all of them and then there are a bunch of intermediate species and the plot looks similar to to the plot that we just seen before so basically there is so the distribution of species that then can be identified as habitat generalist if they are found in every single resource or habitat specialist if they are found in one or few resources kind of mirrors the distribution of metabolites and this was our idea let's try to understand ah wait I have another slide before going there and let's see what time is it plenty of time what happens if we ask this if we see how this generalist and specialist behave in the moment we look at multiple resources so here is the same plot of the richness increasing with the number of carbon sources and you can see that this generalist species that we see in single resources kind of remain the same across all the combinations so there is a core number of species that are always there no matter the combination of resources which resources are present and then there are and actually the increase in diversity depends on these specialists that increase as we provide more resources in the media and of course you have also a gray bar because since we are starting from let's say a big inoculum you can get differences so the gray species are actually the species that were not present in single resources but then appeared when we looked into multi resource environments yes what you think to see these gray species do you think this is a sampling thing or do you think maybe this has to do with concentrations or something like that for sure there is the fact that so when we start these experiments you have this inoculum and so is not exactly the same species that you are inoculating in each well so each well has a sample of species that belongs to the same pool but might be different the other thing is that it is possible that if you have more resources you can actually support you can produce a slightly different metabolic pool so you can get actually new species that you couldn't support before so I think it's both but I'm more inclined to say there is a huge let's say effect of the pool that is the same but then you get different inocula two, okay so if I understand correctly you define like a specialist in a generalist depending on how many resources I mean when you supply different resources like in how many resources you see that single species but then in my head when I hear specialist in generalist it means that species can grow on that specific resources so how do you know because for example if you see a species in the starch community in the cellulose community how do you know if they can actually eat the cellulose on their own or they are like eating something that is being produced in the metabolic cascade let's say that comes from the degradation of it yeah so fair point I'm calling them where is but in the paper called Habitat Generalist and Specialist because I don't exactly know about their consumption patterns so what I can tell you is that I know afterwards now I'm doing experiments with isolates and I know that there is a fair amount of species that we find in the communities that actually cannot grow by themselves in the carbon source where we find them and doesn't mean that they cannot consume the carbon I think I heard you because also I don't know how many but since we find them I think they can grow fast enough so that I can see them but maybe they are missing a cofactor that someone else is providing an amino acid that someone else is providing a vitamin so there might be several reasons for them not to be able to grow but from this data here is just let's say Habitat you can call them Cosmopolitan or what's the opposite well so let's say they generally start the Cosmopolitan especially start the rare species alright so the picture that we're trying to build here is that in our communities given what I just told you so the previous six slides our idea is that these habitat specialists are actually these sorry microbes that tend to prefer glycolytic substrates and might actually prefer this glycolytic direction of the central carbon metabolism while these habitat generalists that we see everywhere might be species might be species that instead can be everywhere but prefer the gluconeogenesis direction of the central carbon metabolism and mainly feed on organic acids the point is since in our picture organic acids can be produced starting from sugars as well then basically everyone in our community preferentially consumes one type of substrate but since organic acids are everywhere these species can be generalists this is our picture for our view of our communities from the data that I showed you before based on this we said okay let's try to build the model so I'm showing you again this map which is the object of problems but let's say we decided to use a consumer resource model because we are in a resource environment so let's forget about Lotka Volterra no phenomenological model today but consumer resource models what you do usually is that for example what Fankaj does Fankaj meta and he presented this model is okay in the consumer resource model I have the dynamics for the species I have the dynamics for the resources and then I have a matrix that describes which species consume what resource and what they leak in the environment and there are different ways of populating this matrix one simple way from a statistical physics point of view is random so we decided instead to use the information that we got from this matrix and actually all the ideas that I showed you before that everyone is a specialist for something but for different things to populate this matrix and the assumption that we made is that again everyone consume preferentially one resource and leaks in the environment the resources that are adjacent to the consumed resource in our metabolic map and so on and so forth until you basically reached the bottom of the food chain and as a side note you can see that the concentration of for conservation of mass if you start with one concentration of the resource the concentration of each metabolite that is leaked in the chain must decrease and decreases as one over R so if we do the simulation keeping this in mind we finally found that we can kind of recapitulate this relationship between the number of resources and the richness that we see in the experiments and the funny thing is that we kind of also recover this distribution of habitat generalist and specialist that we observed in our data and yes what fraction of the nutrient leaks after consuming the primary resource I don't remember the value because this is kind of a sensitive parameter even if you look at this if it's too little well I'm pretty sure you use pretty high because otherwise cascades would not be possible it's not, yes I don't remember the value but it's pretty high because in the end you have to allow the production of it's a long chain that is related to how we constructed the map and I think we we struggled a lot to try to get the model that was doing all the things that we see in our experiments and we're still not there because yes we get some variability in single resources but not much then we don't really see the averaging effect to resources so the mainly result that we're able to recover with this model is actually the linearity in the relationship and a bit this generally specialist stuff but it's kind of the built in matrix let's say is crucial to get this generally specialist map so what I'm trying to say is that I really think this model is wrong for many reasons which is particularly wrong because as I told you before the other question is how diversity changes with resource concentration so we have a third axis that we need to explore well maybe not today but just a sneak peak so why this model is fundamentally wrong because it predicts that the diversity of community should increase with the concentration of resources and this is for a simple reason so we said we have this chain that I showed you before so the concentration of the metabolize decreases as we go towards the end of the chain so there will be concentration threshold at which the concentration of the metabolize is not able to support anymore the growth on one species this species is excluded this is what sets diversity if I increase the concentration what happens is that I have more let's say more stuff that can be converted into metabolize the chain is longer diversity increases so this is a very strong prediction I think that when we were looking at this model yeah we have this prediction it will never be true and so we said what we need to do the experiments to actually verify that it's not true here in one minute is an experiment that I did again with starting with several carbon sources and I said okay let's focus on let's avoid polysaccharides let's just focus on sugars and actually I have disaccharide and I try saccharide the organic acid and some amino acids I give them a different concentration and actually at four orders of magnitude change in the concentration I always start with my soil soil communities and I do the usual deletion protocol still shaking a lot and what we get is that well for sure diversity is not increasing with resource concentration so here it decided to divide between sugars and the rest so with sugars you can see we have four orders of magnitude change in concentration the richness is slightly lower compared to the previous experiments but I don't have polysaccharides here that kind of rise a bit the average but you can see that well diversity doesn't really change and if does something probably decreases with the concentration but here I'm plotting the slope so if I look at this individual slopes of each resource as a function of concentration is not spread out but none of them is actually significant so none of this relationship with concentration is significant and is it the same if I go to organic acids the answer is yes and here I'm skipping the highest concentration because I discovered that M9 buffer is not able to buffer the acidification of the environment anymore and so pH is completely wrong but you can see that basically diversity doesn't really change and if it does something is a bit decreasing and this is again the same plot with the slopes for different resources of each point so the region has a function of concentration so not a big change but the question is maybe at least structure is changing well I'm just showing you here glucose and malate well for sure there is these three concentration in these organic acids where there is mostly one family or order dominating structure doesn't change that much we might see some change in the in glucose as you increase the concentration but still there is a lot of variation so it's even it's actually difficult to decide whether to for sure this you don't see many changes but here is also difficult to pinpoint how the structure is changing so I can't go into this but so I will just so what I'm doing now is actually trying to understand from a bottom up point of view if at least I see changes in the structure of communities and whether I can find any pattern that is related to some sort of general metabolic principle that can explain these changes yes so in the one percent it's close to one and the for the different concentrations so one percent is one and then it's tenfold less so it's like point one so it follows a bit in the community it follows really well the so the carrying capacity of the community can be retrieved from the total concentration of carbon very well actually if you go into single species no so if you plot the logarithm of the concentration and the logarithm of the OD they are on a straight line very high so takeaways I think that when I started going to resources while I found a lot of surprises there is a violation of competitive exclusion no effect of concentration that is where all things are kind of expected and the other thing is that I kind of believe that life strategy is linked to the preferred direction of the central carbon metabolism can also help us understand a bit more of how these communities assemble not just the physiology of single species and well there is as Terry pointed out yesterday there is a mismatch in what people measure when they look at what metabolites are secreted with actually the number of coexistent species that we see and this is kind of a mystery that we haven't really solved yet and the other point is that the state of the our consumer resource models they are bad they are wrong they are not really helping us understanding what is going on in communities so there is a lot of work to do especially for theorists so and with this I didn't thank all the people that helped me with this work this is my PI at MIT Akshit and Heanne were my collaborators in the resource papers we are still working on this Claire, Karine and Yaron are my collaborators in the temperature project Jana, I am working on with her on the saline project and this is our lab that well this was a picture taken while we were moving so you can see the state the disaster thank you for having me of course late, but I think we have time for a couple of questions hi, thank you for this talk I had a question on the experimental setup so as far as I understood you had this 96 well played with different media, with different number of resources but the inoculum always came from this soil community so when you were saying for example when you were going from one single resources to the two and you were saying maybe we will see the union but you actually saw the mean of the species did you also try to mix the communities that had established in the single resources do you expect that you would have the same if you did because I am not sure then if I understood when you showed the plot of number of resources that you said like we can go back and forth like if I have a community that grows with eight resources and I remove four resources what will I see the same richness? no, ok the experiment the way you describe it is actually how I did them so there is no mixing later but then it becomes a matter of coalescence I think it's a whole different ballpark I think also so Alvaro was exploring this kind of things so I didn't do that and what I meant to say back and forth is that you can let's say establish a tournament so if I mix these two resources and you can try to ask every time ok, can I predict based on which resources the constituent resources can I predict something from the communities the constituents with the constituent resources or about the mix thank you ok, so I have a question on the last part of your talk when you were explaining that crossfitting in a linear chain so you increase the initial concentration you would increase the final concentration of the last resource of this chain so you can increase the richness in this sense my question is when you are changing the initial concentration experiments how are you sure that you already have a community of species that are really able to consume at least one metabolite more in your chain if you don't know what's the community how are you sure that you're really changing concentrations and really changing richness no, no, no I'm not sure this is just to we should use the consumer resource model we can assume in the same thing even if we don't know so when you try to ask with this model how is consent so it's just how the model would work and why is predicting this thing this is just a cartoon to highlight ok this is why this model predicts that diversity increases with concentration and actually why it's not working can be because not everyone well, you have to also understand you don't actually know what's around so what they are leaking and this was just to illustrate how the model in a very simple way how the model works in terms of predicting diversity from changing concentrations this is all very interesting did you try to see the effect of the dilution factor just in the spirit of your first talk here the death should help increase in the dilution which is kind of equivalent to death here should increase influence of the of the fast growers so it could be interesting experiment to follow up and what was the dilution factor you may have mentioned it 30x what factor did you 30x it's a matter of convenience because I have these deep well plates of 500 in the bell so I work at 300 ml so I can transfer 10 ml and I don't have to change the head of the 30x I don't have to change the head of the biopro so so here's a question I had actually the first time I heard this growth dilution the first time ever and I still have this question and I'm going to ask it's related to what Sergey is asking so I never thought in principle this diversity will be issue in this experiment because some cell die and thousands of metabolites release in 24 or whatever how do you respond to that so in 24 hours there are not many cells that are dying what? Alvaro tested it Alvaro I think had this test in his paper and they have 48 hours so they stained the dead cells in their communities and they saw that there was not that much death but the dead cells could all be eaten up yeah okay but then it's like right how do I test it no my assumption is that they're not dying that much in 24 hours some of our cells die was like a half an hour then I ask you a different question so from my experiments to Alvaro's experiments I have more diversity than he has in 24 hours to 48 hours in principle they are dying more over 48 hours it should seem more diversity than me we don't understand no no I'm trying to we understand little about growth I agree with you but from what I know about death it takes a while for them to die it takes a while for them to die no necessarily depends on strength depends on conditions but what I found out about death is that I cannot exclude phages so if there is a phage like burst because I'm taking them from soil they can feast but I think there should be more random path then more should be more random changes in diversity if you observe in the middle but if you observe at the end I mean that that one has died and they have recycled eventually but don't you think that then there should be more so when I repeat these experiments they are kind of they are pretty consistent if it would be just a matter of dying of phages and burst should it be more random death is very regular just because we don't study it no I think that where actually started but I think this is actually embracing I think that death is a problem it's not just regular death it's more phage growth dilution cycle growth period so then you hardly have anybody dying because if they are all growing if you observe the OD it settles down after the moment no but one thing that I want to do is actually change the dilution period but the concentration the very low concentration you are looking at that's very little nutrient that's like sub millimolar maybe 100 micromolar type of nutrient but they are far away from but they are far away from the K no but then they are just sitting in stationary phase longer well if you measure growth rate we can I show you the curves but they grow a bit they have to grow a bit grow a little bit then they are sitting there that happens when they are sitting there we can look at the curves happily ok, I think it's a good moment to stop I mean it's not a good moment to stop we could go ahead for hours and it would be fun but we don't have a limited time