 is again Terry. He needs no introduction and he'll tell us about dynamics. Yeah, so first I'll tell you about a project where we are working with a simple interacting pair of species, okay, but a physiological study of what's happening in that system. Then in the second half, yeah, then in the second half I would like to then go use a lesson learned there and describe a different model of community dynamics and with a focus on getting a model with a few parameters that can generate predictions. All right, so yes, okay, so we already talked about species diversity and this landmark study was already mentioned a few times. It was really, that's what I worked last time, again. Right, so there was this landmark study by our Rosentius lab and together with Pankaj Mehta theorists that got many people's attention including ours. So they took, I guess, community bacteria from the wild, from the prance leaf in this case and did the type of a gross dilution study that Matino is describing and the key to this experiment is that they feed these bacteria in a very simple medium, just glucose is a carbon source, right, and then after a number of these gross dilution cycles you get to a stable community at least by kind of a composition but it's more than one, okay, and that's surprising because it violates the principles of a competitive exclusion where we think one species per niche and for this type of simple problem one thinks about nutrient niches, right, and so now the kind of explanations for the existence of this diverse community is the cross feeding. Okay, so this was actually as told by me was Pankaj Mehta was the theorist was actually driving this project. Nobody would even do such an experiment biased by this competitive exclusion principle. Why would even grow a community with a single nutrient source, okay, and from his theoretical study he thought that might be possible and indeed it was possible. So anyway, that was described to me from Pankaj and so this is a typical type of a model, so you have a nutrient that's consumed by a number of species, in this case we have just one nutrient of glucose and then these species excrete the metabolites and then metabolite servers and nutrients for other species, okay, and the so coexistence possible even in steady state, so these models were solved in steady state. Okay, now I look at this kind of explanation, rationalization, and I was not convinced by it based on what I know about bacterial physiology and the reason is a matter of a number. Okay, so in order for excretion to support to sustain a diverse community, diverse community meaning that there's at least 10% of biomass that's something else that's not you, right, I mean more likely we're talking about 50 maybe 70%, one species probably not more than dominant then, well I mean there need to be macroscopic presence of a number of species in order to be a diverse community and for that to happen the guy who's primarily taking the nutrients will need to excrete basically all of half of what it takes in and that's a lot to ask for for somebody that's growing in steady state. Okay, and right, so typically it takes 30 or 40 millimolar of a carbon to to to carbon atoms to generate one OD of cells and one OD is I say 10 to the nine cells per millilitre, okay, and then that means that you need to excrete stuff at the order of a millimolar per OD, okay, but when I look at the actual excretion, like the measurements that were done, there's typically a hundred times lower, okay, and so there's some so we went through the literature there was some very well cited examples that say happening in the ocean oh so much of CO2 like 80% of the CO2 is being excreted or something like that, okay, but when you look into these experiments the excretion of carbon positive during growth and limited growth was actually quite moderate, okay, and a big excretion comes after they ran out of nitrogen or something else, okay, but that's not surprising if you're limited surely you can you can dump with the unlimited supply and that the unlimited supply that's produced will not help you in community diversity because you're limited by nitrogen, okay, so then so I so so I have this conundrum in my mind how is that possible I so I wasn't quite believing in this excretion just it's a very kind of a communistic type of a description right and everybody's helping each other or great but how do you prevent somebody just why would why would cells do that when they're basically happily growing by themselves exponentially, yeah. For anaerobic growths would you expect to have the sizable excretion of carbon when you're limited by energy? Yes, I get to that so there's some exceptions right and the so there's of course the synthetic stuff that you know you micro cannot grow by themselves they can only grow by the cross-feeding product okay but we're talking about natural free-living bacteria right the we have the supplementation process during anaerobiosis right so without oxygen organism invariably is free lots of stuff but this lots of stuff you know when organisms cannot you most likely another organism cannot use either and they're even toxic to most people okay then it takes a very special collection of a bacteria that can actually take these what would be normally poisonous to others and to use them and then then it takes the next one to use that eventually pushing the flux all the way through only you get to CO2 and methane then then they're out of the system okay if you don't get there it stops somewhere in between and product builds up everybody's dead right so so it's very difficult so we're not talking about that kind of processes okay and aerobic fermentation because minimum you know that's a E. coli bacillus many organisms they excrete acetate right when growing fast okay and this the amount that's excreted is almost say 10% okay so we're getting maybe a little bit into this number range but still not a lot okay so if you just excrete 10% I mean it's not that much and and what's excreted for very special reason and and very defined product okay so so then we decided to take a look at what's going on well this was a on the one hand we're thinking about these complicated communities the reading and the kind of the China digest the result but on the other hand we was very sort of grateful that a group of ecologists that invited me to join this collaboration the headed by Otto and I had no experience in ecology right so then then we start to learn about things outside of E. coli and I said well let's just play with them just put two strings together and see what happens right and so this is the work I'm gonna be describing by a former postdoc Kapil Amanes and they published just earlier this year right so we took the system that the Otto described that the cutting bees and the collecting marine bacteria on cotton bees and there's a succession dynamics that happens as a function of time so various bacteria get on to them he said yesterday mentioned that initially there's a wave of a bacteria that lands at their chitin degraders and then other stuff coming right so we took two species one is a various species that degrades chitin and then there's another species that happens later we don't know what this is what is take them and just put them together and see what happens okay and so the we grew them on chitin so the severe species attack and degrader grow as advertised and the other species bee species did not grow and then we find the same thing for Gluknack for growth on Gluknack which is a monomer of a chitin okay so then we forget about chitin was just look at the monomer right and so if you look at the growth of this a species video on chitin alone you see that the blue line is a Gluknack so Gluknack is disappearing give you the yield of a Gluknack acetate is being excreted and ammonia is being excreted okay and that's easy to understand if you look at the molecule of a Gluknack it's got a acetyl group on it it's got a main group on it and it's a group cause otherwise okay so basically it kicks up the acetyl group it takes some ammonia for itself and it release the remaining ammonia okay now we take the acetate group the acetate and ammonia and a feeded 3b05 on the individual aid on B the bee grow but actually a does not grow so so the A species cannot grow on acetate alone yeah when you say it excretes acetate and ammonium these are the two products that are mostly excreted but do they do they do they excrete something else today what excrete something else very little I mean just in this phase okay as far as we measure we don't see much yeah so with metabolomics or you actually we did my topomics yeah you know you always see something but nothing I'm always talking about flux how much flux it becomes a noticeable okay thanks yeah and yeah so actually the amount of acetate that's excreted it's basically the sum of what you expect out of our acetate overflow metabolism plus the amount of every every molecule it gets out as a plus it excrete something little bit more yeah okay so that's the base state we are talking about and this leads to a very simple scenario of what was so we're just commensurability right so B is basically just taking the waste part that the excretion product of a right now if it in the medium with a weak buffer so for example you know we use the ocean what a buffer of a two millimolar of a bicarbonate which is just what's fixed by the equilibration with the CO2 in our atmosphere okay then as a as you dump on the order of a two millimolar then pH drops okay and then so there's a toxic effect 101 and a great that's really if I can take it so now this becomes a standard scenario of a simplest example of a centripetal right so one is taking a waste removing a waste product okay and okay so so now for this classical scenario immediately what one has in mind is that a there will be a crossing point with a stable coexisting of the two strain right as long as the effect of the waste product is a has a stronger derritorious effect on the producer than on the consume right so so that you have these two curve crossing well this turn on that to be the case so when you vary pH by itself you look at the individual organism actually the the consumer is trivial five which it grows on acetate right but then it's actually more sensitive to pH then as a result the standard scenario does not occur okay and so meaning this acid has actually stronger effect on be there now okay so this was a really a shock because maybe biased in our own ways I'm just seeking for mathematical simpler just assume that the first scenario is always gonna happen okay but actually there's a 50-50 chance that you know the second scenario will happen okay you produce them a toxic stuff so why should the always the producer be hit more than the consumer sorry just a clarification so with the strong buffer do you have coexistence based on the growth with a strong buffer they just grow by themselves right so explanation until and to everything is used up but and they could exist because the second one grows on acetate huh they coexist when growing together with a strong with a strong buffer in batch culture course there's no issue of a coexistence or not I'm a day if you were doing if you were doing serial dilution with serial dilution they coexist yeah actually I'm not showing you data but it's a it's in a paper yeah with strong buffer you do serial dilution they just coexist all right so then okay so that what's what's going on in this case so so yeah so there's half of the scenario basically was a brand we do not consider okay the so what what's happening there and then you think about it could because a always grows faster is growing on glucose I mean just faster than growing on acid okay and a produces acetate right so you know always I think it also takes one to get all the well I mean because exponential growth is nothing you can you can be okay and so if it's more toxic it's it's more sensitive to us yeah if these most sensitive that they're just gonna crash at some point okay then you can predict at which point it will crash because we know the yields and everything the excretion and indeed it crashes at some point as it builds up and then it's out the system is out of business okay and if you look at here the gross curve for the top figure and you see indeed it crashes the culture crashes that's where pH drops the orange curve all right and then if you look at the group neck acetate so pH drop will correspond to the acid arising okay and then group neck was only like half consumed at that point okay so the system just freezes the so the so yeah the classical centrifuge scenario does not apply right and if you look into individual species so the OD just tell you the sum total of the biomass you see that actually the a species dies very quickly okay it crashes as within a within a day it reduces three orders of magnitude yeah yes very good we'll get to that yes it kind of free the OD freezes something's still happening and that's key to do well okay right okay so so basically it freezes and and a Vibrio just dies very rapidly under under this low pH in a low pH the situation okay so then we do the 24 hour gross dilution cycle right and we do not certain this kind of a system we do not expect coexistence and of course we get coexistence so well what actually happened was we first got a coexistence right and we try to understand we thought to the standard scenario and it's not a standard scenario okay you do this for high buffer you get coexistence nothing fishy but with the weak buffer you get into this weird situation so so the first after first cycle first day indeed there's a three-order drop of the species okay but then after three days they they get into equilibrium under the fire is it clear all right all right so I said they do coexist okay and if you look at the group not gonna acetate you see that after three days when they get into a stable cycle like the system was able to consume or cover so and immediately our our friends were challenging as well are you sure you this is not mutation right the strings are changing or yes we're sure that because it could repeat this process take take the product after five five cycles to do exact same thing okay all right so the but then now we say well what's going on and then just to check we change the dilution time to six hours okay so in six hours we choose six hours because during the first six hours everything is financial growth okay given that the initial inoculation of the we are using and so nothing fishy should be going on and indeed if we just have yeah so if we do a very rapid dilution then the only only only a survive speed doesn't just be kind of keep up with it okay so what this tells us that there's some kind of a weird thing that's happening in the stationary phase outside of a six hour growth phase okay that is a making this a coexistence possible yeah question all right so then we look at the dynamics during the cycle okay so we do a number of cycles after five cycles in what we call a stable cycle and then we did measurements in between okay and you see some very interesting behavior okay so this is for the ex dilution is doing in our experiment right you see that first of all as basically the system stops growth stops after about half a day and then after that there's no death so in a stable cycle there's no death even though in the first cycle there's massive death okay so it got itself out of the death regime and then the there's also in the initial six hour there's like a warrior ones not growing there's a long lap okay but despite when we're not growing three zero fives growing okay so so B is supposed to rely on a to make stuff that to grow okay but it is not growing but B is growing all right so as and then we when we make a measure of the acetate in a gluconectin we see that around 12 hours when things stop like the acetate peaks and then rapidly drops so now the rapid drop explains the lack of deaths right so the lack of death is caused by drop in pH and if acetate is gone then there'll be no death okay and okay so then the interesting thing is of course what happens in the speaking in the brief so several hours regime where acetate is rising and then dropping okay so then we zoom zoom into this regime and you see that in this regime actually so as acetate is created three bills three bill five slow down right and then then actually picks up again okay and you know before that and it's actually growing on a faster rate in this regime and this faster increase we presume it's what's responsible for acetate disappearing and okay so then we measure okay so first by in our lab by HPLC and we find the clear excretion of us the major excretion at the order of a millimeter per OD right of a pyruvate and a lactate and some glutamine other stuff just just during that window and and then with with ours that's a help and then we also did the metablonics and see all kinds of things that's excreted during the window not before right there's a very little before during this window okay so what's going on so then we did the whole bunch of experiments to sort of get to a picture but I'm not gonna give the give you the process of how we got to the picture I'm just going to tell you eventually what happens okay and I'm gonna build in basic first of all address the question well what happens when acetate build the pH drop why cells stop growing okay and that was actually a study we first did on E. coli a parallel project that was going on in the lab okay so we have a Glock neck being consumed by cells and the the yeah so it goes through the usual basic glucose and it goes through the usual glycolytic stuff and then excrete it kicks out the acid group and it's a form of acidic acid acidic acid is a small neutral molecule it can cross into the membrane okay so acidic acid concentrate inside and outside is the same now the acidic acid of course in each compartment is in coexistence with the acetate anion and there was a proton and I ended the the the partition given by the local pH okay so then there's actually a very striking result due to Russell many years ago that if you take a ratio of the acetate concentration in the two regime well they're just given by the difference of a pH right because the HAC the acidic acid concentration is the same okay now if you have a pH 5 outside and pH 7 outside the ratio is a hundred okay so that means if you have a 3 millimolar acid outside that's where the acid crashes you have 300 millimolar inside okay now you should you're supposed to be shocked by 300 millimolar of stuff inside cells okay because you know the typical the similarity of the system is is a 450 millimolar and you're basically a appreciable part of the it's okay so in these cells normally they they balance our similarity by producing glutamate some other awesome lights in the cell okay but then when you shove that much acetate into the cell and there's nothing they can do about it it's just a physical chemistry right then well what do they do the most money still has to be balanced right so then what they do is they're lower the all of the metabolites inside inside the cell okay so lots of stuff got kicked out but the glutamate aspartate you see a product that they all get kicked out right and I just show you one piece of measurement but then we did extensive metabolomic study in in this paper on the system and what you see is this is total metabolite of the cell with the norm without acetate this is this is data for E. coli similar type of things going on for one year one and then with with this 30 millimolar of acetate in the low pH then you see acetate replaces half of the internal metabolites okay and so yeah so this plot shows that as you at a fixed external pH as acetate concentration increase then some of the endogenous of a decrease and this increase in a way that the total is balanced okay so it's like it's like basically you're filling the cell with useless metabolites and kicking out the useful metabolites and that's what growth decreases yeah during this transient because of these osmotic processes you might see dramatic cell size changes maybe yeah I mean the transing everything so this that is doing steady state right and so yeah but then during transing of course where you have an osmotic shock basically yeah during osmotic shock all kind of yeah but this is kind of a it's like yeah hypo osmotic shock you have lots of stuff and the stuff will get kicked out so within 15 second gutamase are being kicked out okay but then they reach a new steady state where production of all these metabolites are reduced so that they don't in the excreting the long-term okay so to keep this okay all right so another thing is there was a lot of acetate coming in to grab all of the CoA okay so that so that normal ecolysis cannot go down from pyruvate to down to a TCA second right and the cell then the cell continue to take a group neck and excrete pyruvate out okay lactate is a variable it's needed to balance reduction power okay and so that's what ecolysis doing that's what we're always doing why does it do that presumably we say that it's it's using it to generate some ATP to do to fight to keep things going okay and then this also explains that when when the carbon source is gone then it just dies rapidly okay so it's it's it's so it's continuing excreting stuff in this case left in a pyruvate right to get some so to get some energy out they cannot use TCA okay because the TCA requires the TCA intermediates which are basically wiped out by by this thing that hits it okay okay yeah and yeah so and then this is a intrinsic effect of a week as you can try different butyrate papaya may whatever they did they have this type of effects all right so what what's happening on a 3d o5 the side of the consumer 3d o5 okay so a normal pH there anyway acetate eaters so acetate comes in in a normal way right and then they gets incorporated into the TCA cycle and it's through the carcass very shunt they go up through the gluconeogenesis right now at low pH again same thing happens the acetate overwhelms everybody right and the stuff gets kicked out and you can see that the 3d o5 this is by itself excrete glutamate aspartate in the stuff right until several hours later basically when all the products are drained there's no more to excrete the reaction starts okay and without a TCA intermediates of course gluconeogenesis doesn't go and I just come to a standstill right so now you can see what happens when you put the two together okay in a one case the one is producing part of it in the lactate which we'll just call it Parovay right and then so Parovay could be taken up by 3d o5 right and what is taking up and they can do and they can go to places it can go up and it can go down okay and when Parovay refills a TCA pool then the we can have a we can rescue the TCA cycle and once that's rescued well then there's a way to get rid of acetate so then we can basically it's a way to burn acetate into CO2 right and so then eventually acetate gets burned off and okay when you see that just burning off acetate is not enough because then on the other side gluconegesis still a producing acetate you still you can still have this problem okay but now in a stable cycle the timing has been arranged such that just as 3d o5 is recovering burning off the acetate the gluconegesis gone okay so the delicate balance that's that's that's sort of a reach by the system autonomously so at that point then there's no more production and and just everything comes to a standstill and they get ready for the next cycle okay now when the ready is 3d o5 ready to go immediately okay when gluconegesis resupply when one cannot grow because it's depleted I mean many of its metabolites are depleted okay so there's a long lag right and when gluconegesis comes in it's still processed as Parovay gets the energy and it gets us 3d o5 to grow but then it takes a while for itself to recover before you grow again so that the picture basically explains all of the features we're seeing okay so there's a there are many well three years of painful experiments that led to this and I'm not telling you about that details of experiments like so it's described in the paper okay so then and with this understanding we can build a simple model but there's a model with actually quite a few components but it's just standard resource allocation models and the consumer resource models okay but we need to introduce several phases describe each of this phase there's an exponential growth there's a gross arrest when acetate builds up right and when when there's a when the lots of acetate is excreted right so you can it can you can cause gross rep by 3d o5 or 3d o5 is that it can be kicked into the gross mode and eventually that's it can cycle okay this is a well we have a most of the parameters because we measure we try to measure everything right and this this system the it's a complicated system with but pretty much all parameter fixed by experiment right and we're able to reproduce the dynamics this is at the end of the cycle so you see what what happens for the two densities and the concentration which basically recapitulates the observation but it also allows you to get a peak of how this happens as it goes into the stable cycles okay and you see you see after first cycle it crashes and we will have to include that and all that stuff okay and then we'll include everything and we just reproduces what we observe okay all right so the now this is about a system isolated from the from the marine world right so then we we since now we understand what's what's underlying the system so well maybe this is more general than that right as soon as you have you have two so what we have a system of one sugar eater one acid right to the two opponent system that I will talk about yesterday right then we preserve them a generically this should happen right so we tested this so here we took another pair of the from this is a pair of a strange from a from a soil bacteria isolate yeah so sometimes like these two species they both can grow on glucose and yeah yeah so this what we're seeing is in this case is so one person when you have both sugar and acid right some species prefer sugar some preferred species prefer acid right but in our case right now not this case I understand yeah yeah but the other case is you can grow on both yeah so so so so we look at other cases right yeah so then here is one case I have no idea what it I know well I know so the mono cell fluorescence grows on both glucose and and acid okay although it prefers acid right and so we do the same kind of things the this is a monoculture growth right so they they both grow fine for for a while actually at the similar rates right and if you do this growth dilution cycles they reach a steady state a stable cycle right and if you look at a pH that's very easy to measure and you see this a dip right which is told is to know we got the strains from our overalls are correction and we just took so we got some one that prefer glucose or some preferred acid and just get them to the again okay and and if you look into what's happening you see the excretion of pyruvate and acetate stuff during the cycle yeah if this shock is due to this osmotic internal problem if I do the experiment in in hypereosmotic medium do I delete the do you expect to delete the so this is in hypereosmotic medium already right so this is in hyper hyper yeah so so so the same phenomenon because it this number goes so high but it will shift that if you change the the outside of similarity it will shift the point where this happens right but it happens anyway yeah so we did that for E. coli not not for Vibrio okay yeah it shifts the number all right and then so so so this then then okay so here's another try to know just take E. coli and as soon as potato they have no chance of a being together right and we fed them over the glucose like the way in the in in the in the experiment okay and we and though and we look at the way they were growing cells my in in these place without shaking and you see that okay the singles strain growth like this and they again they get into stable cycles okay and during the cycle you have this step okay and if you look at what's happening inside the dip and you see the excretion of various okay just I mean I think I understand the mechanism but there's something I don't understand when you say like in our experiments in the sense that if I understand correctly they perform experiment with strong buffer right so you should expect that in that case you don't have these so so so let me yeah I was too fast so in this experiment right we we use we grew neck on them so we took this strain from from from a soil but then maybe they have no business was looking at right but we still use a group not cooking against the kind of a nice easy way of dumping acetate into the intermediate and they see this happening okay and then in this experiment right what we did was a feed them with glucose okay and with with the strong buffer okay but the thing is they were doing experiment without shaking without shaking means you get into anaerobic growth okay and yeah so then if we if we do the shaking all this disappears okay so during anaerobic growth so much stuff much more massive excretions possible that doesn't matter how much buffer you put it's gonna break your buffer well of course you can't know put an unreasonable amount of buffer than you have an ordinary problem yeah yeah so can you help me understand my experiments I do them with shaking and I see coexistence and it's a strong buffer yeah well I mean I don't know I'm just I'm just talking about right in the script so because I remember that in the in their paper they actually I think one thing that the the reviewers asked was exactly oh but you're not shaking or do you have do you have right that's the pH that their measurement right so P you'll get a pH 6 with the amount the kind of acid that you have this is gonna happen no no but I remember that they made a point about the fact that oxygen was they didn't have any stratification with oxygen you did they have some oxygen when you're shaking have something okay but if you do with this effect it doesn't require oxygen being zero it's just that we just look at a pH I don't know for whatever reason pH is right the pH is not very high and we're very yeah it's a kind of a six ish right so between six and a no you call it a normal pH is seven point something right so there's at least a fact of ten different so when with excretion you easily get to a tens of millimolar and you're getting to this kind of problem you have a big multiplication fact all right so so summarized right so we have this unexpected kind of a kind of a metabolic exchange yes so cross feeding ultimately if you step away so yes it is cross feeding that rescue the things okay but it's not the kind of the that's imagining in MacArthur's model right I mean it's it all depends on dynamics and and all that stuff right and it kind of resolves the conundrum I had in the beginning that why should cells be excreting lots of stuff when they're happily growing whether you know the only when they're stuck for some reason in this case for acetate right that they start excreting stuff okay they cannot use this stuff anyway they cannot to they cannot grow they cannot use it and the time for it out maybe somebody else could make use of it and the rescue hey Terry can you help me explain when you showed the results in the very beginning it took multiple cycles for this to happen so why does this phenomenon not happen in the first cycle of growth between you know the the degrader and the scavenger what I mean the first time you know when you when you introduced this in the very beginning you said that in the first crash at the first cycle it crashes yeah so so this is actually then we actually look into our model to see what's going on right and because we trust our model enough that it can give us and then we can do experiments to measure right and so you can look into this model right so the first cycle in these crashes several orders of magnitude okay and then you start the second cycle right so then the now now a takes it takes a lot much longer right before you can build up it was just kind of a this kind of a detail I mean eventually you know it's just yes the you know you have a stability mechanism eventually the diamond find itself to the fixed point yeah nice control would be to now that you know this the steady state ratio between ones or at least the start in gross ratio if you now prepare your inoculum at the first cycle in what would be eventually as a steady state then it should establish from the first so we did exactly that experiment yeah right so that is just first few cycles one or two cycles they negotiate the depletion times until they get the steady state yeah I know it is I'm trying to cut a full talking to a half so skipping some corners yeah but but me if you read a paper me okay yeah but for for some reason our reviewers really did not do not like this work or just you're talking about this cross-feeding we all understand that okay all right so so yeah so this is so this is a so 101 right so for in this particular system the it by excreting this other metabolite during during the time when it's stuck it's actually helping itself also at the end is rescue itself on death right of course not to mention the keep things going all right so key requirement here is the complementary models essential metabolism this is not to say you pick any two strains you get it right but in the environment there should be enough of a one type and the other type so it's something I can count on there will be some assay either out there right under the chance that they could they could do a rescue me right and some general lessons right I guess number one is this is a coin think about coexistence not as a steady state it's just DDT go to zero right but as a stable limit cycle yeah so even though everything looks fine right like critical thing is a happening just like in a two-hour window everything is happening that to our window okay and then you otherwise you completely miss it right so suggesting experiment we really need to do dance sampling right to to catch these moments and then the number of lessons for theory right so this is a simple system you can get right just because we really took it like I said well you know was a E. coli was so my lab started with studying of the lab promoter okay and it took us several years to study the lab promoter this is basically okay look at the simplest cross feeding right we learn something and indeed we learn something not sure you know this is a red herring but then we would say at least in several cases right it also happened that a system not saying that applies to everything but no it happens right and yeah so and maybe the biggest lesson we get out of this is that the complex behavior in the community right can come from the combination of internal states okay so each each of these speech you have only two species each has several state right and and all of these companies to happen is an interaction of these states okay so so so the several variables within already one organism it's not just growth it's not just a density okay so that is hidden variable right so the in principle this hidden variable could contribute to the complexity of the system okay so that was that was the lesson and yeah so let me wrap up but yeah this part right so yeah couple the hero of this work and our nation now as a theorist I actually I did not think there's a chance to describe a complex isn't that that was actually but then I guess we have enough parameters so that we put it in it actually just just work together okay roughly will yeah and we saw it's lab for metabolomics and without auto none of this would have ever happened so yeah so very it's very happy so this is our first kind of a ecology project that dipping a total thing to into the world of interacting bacteria so now I'm gonna switch gear and talk about some elaboration of the ideas that learn from this okay because one well first time I talked about this was that a KITP and basically the message people got is oh it's a really complicated Terry look at a simple system and it's a really complicated right and everybody's pressing me can you distill some more lessons out of this okay okay describe this in some simple way and we're talking as a challenge right and our being Morgan was at the KITP program together and we discussed a lot and then we we sort of a cooked up to something else and then which is so I would say this is inspiration based on this concrete project right so again I will say that okay so the if you look into the dynamics of this so now I'm looking using the model the avenues constructed for the system right as the the actual system is the we cannot possibly measure have the growth rate and the of the two species in every point right but then the model recovers the dynamics pretty well so we're using that we see quite a few regions okay or there's a lab for one or one then the growth and the distance that okay and they come from an interaction with different phases of a of a one-year-old one or three-year-old five no each has like the two phases to three phases and they interact to produce a number of these systems some of the bugs die at some stages so here I'm talking about the stable cycle whether or not I see okay so now then I would say well can we can we distill this into only talk about stable cycle not approach to the cycle okay given that we know there's a stable cycle can we say something kind of a more transparent about this rather than giving all of the details right so then we sort of a looked around and so here's a kind of a illuminated plot so here instead of parting the density we're looking at the nutrients that does being cross-fed everything right so peruvian acetate and a group that and you see kind of a nicer cycling okay and this this plot sort of made us think time they just look like there's some way right to reprimand reprimand rise the system with some variable that's maybe more illuminating okay and we played around again and we come up with basically the two variants of the same thing one you can think about as the amount of a nutrient that's remaining in the system right and that's harder to to measure and the negative of that it's just the amount of biomass okay so biomass kind of a progresses increases throughout these phases and so if we use biomass as a x-axis instead of time right well then we will get the plot like this okay it it kind of a it's like this plot but it emphasized different regime right so for example in this initial weighting regime you hardly see it okay but it's a very critical regime that's a little blip here right it's amplified much more okay so then and they kind of a it's a different way to telling the story okay and so so looking at that as a biomass all right so then we took from there and we built a what we call a community state model right and in this model though so we put it okay so community has a number of states okay that we call the community state right and the organism behaves in a species specific way in the different states okay and then the the community state is assigned depending on the total biomass density that that's the OD of the system okay so that's the that's the x-axis okay so then the we can get into we can we can get into that but no certainly biologically speaking biomass density is a important variable for the system if you pick if I'm gonna pick one variable to get the state to get a sense of the state of the system but biomass will probably on okay any any physiology is the top five choices of something that's indicative right because as biomass increase excretion increase oxygen decrease a lot of things happen okay and now it's also detectable by cells through quorum sensing system like that right but then but we were we're intending this to be is a what example of of a some variable that reflects the state of the system that the cell can respond okay yeah when you have the biomass sort of community biomass as a relevant parameter can you then if I always keep my OD between point two and point four at the way if I always keep keep my community biomass as an OD between like point two and point six I always get one a01 to grow faster yeah so so so so in a model if I if I take this kind of a thing my feeling tomorrow it will produce the dynamics right so it's a kind of a self-consistent way of describing the effort to try to simplify the dynamics okay okay but certainly if I just give some random stuff it's gonna crash right so so I think of as a given that we already know there is a cycle going on right and there's a bet that okay biomass captures this this this progression right then we'll try to see if we can reconstruct the dynamics okay so then in this model we density is only variable okay and yeah no I was wondering if maybe to finally check that the communication that says happening through all acetate and so on we can try to culture the the cells independently but but we controlling the nutrient profile by what the model absolutely this is not a bottom up picture this is so yes that's what we did in the first half when we talk about and all of these things so we so so this is an effort to try to see whether we can we can we can make this system simpler so we're describing away that simple just of course I mean the cell they're not responding in the system we know what they respond to they're not responding to biomass no but what I was saying is like we can try to simulate cells a effects on cells be by us controlling the nutrient condition of say be without say yeah yeah yeah that's what we did that's what we did right but not not here here we expected try to get to a minimal discreet try to you know get rid of as many things as possible we don't want to measure nutrient okay and just to to see at some level could we put together self-consistent picture okay and okay so in this model then each species grow with the rate each species alpha grows with the rate our offer and and it's just dependent on the total it's just responding to total biomass okay and then there's the dilution with the fact of D after when it reaches so we sort of think about gross dilution cycle right and looking for stable cycles and things like that okay so the key is in what is the choice of this function okay so so okay so for some examples so here we actually have measurements say okay so this function like that right the you can use this kind of a this kind of approach this kind of a say stepwise kind of function to describe system like a dioxy shift that's what the Jeff Gorse lab and doing right in the surrogate's been doing that so this already 50 years ago Simon not Simon Bruce Leving has already described by such a picture for two bacteria species on a single nutrient source okay just simple simple way to to describe this okay it's a theoretical I would say this is a simplest way to close the system right you have a you have a growth you have to have you have to have cell density right and then what how do you specify growth rate so then the simplest way to close it is that with with density all right so the in principle you could you can have an arbitrary fun this kind of a functions and this function is used to describe all kinds of a thing it's not necessary nutrient that's the point right nutrient if you have a specific example with dioxy shift so for you can write construct such a function but you can have other stuff that does not allow you to write down in there's no monokinetics right and then we can still just specify growth rate you can still do try to do something like that okay and so so let me describe to you just several progression of this model so in the simplest model we call the diagonal preference model that is so we have a we divide the community into a number of states so here's the interval of a biomass growth from from beginning to to dilution time okay so and normalized to one okay and we divided into a number of regions and so each region has some a width data and index by and and in each region species has a some growth rate okay and in the simplest model to say one regions of one species of good growing okay so this is for one species maybe another species growing region five the species growing region seven so for with some growth rate okay and so then the way to to put all of these information together and in the simplest case the the these sub beans are a uniform if you have ten species and they're just dividing to ten pieces okay and then we just have two growth rate when it's when it's in its own it's growing at the fast rate when it's not in its own it grows us from slowly okay so it's a proud it's a now we have us and so we have an body say and species system with single parameter that just a growth preference ratio of the the growth rate in the favorite state and enough they missed it okay and so what does it do so you can follow you can do it dynamics very easily defines can you can you can do the dynamics and then after a number of cycles then you see that this dynamics it just favors the early species okay so immediately you see a pioneering in fact that in this case only basically first the second species will survive okay there's a yeah so there's a lot of analytics can do this is this kind of system simple enough so you can see how many will survive and can then because well then in the sense obviously the pioneers have unfavorable to unfair advantage so it's growing and it's keep on growing right so then then you can do a case with non-uniform being with or with a non-uniform a grocery preference at the different of these stages and then you can get them into a coexistence all right and so you can you can derive expression for what does it take to get them all into coexistence and their stability you know that you can say quite a bit of us makes quite a bit of statements about about such a systems right so here I'm just emphasizing the early bird effect right that the later species if you want to join a party and be kept in a party you have to contribute a lot if you're coming late right and this tells you quantitatively how much you have to contribute in this simple model right so remember this is now I'm talking about a single parameter right this model was a single parameter right and then it's making some statements about stability right so now you can make it more complicated person some noise right and you see that so now instead of just two numbers I can put 30% noise in the in the favorite grocery in a just favorite grocery for 30% noise in the bean with and so forth and it's pretty the system of pretty stable okay I'm not gonna go through the details but then one thing and okay so this instead of a change in the bean with you can have different grocery biases okay yeah and you get the similar result and so here is part of the fraction of a surviving species as a function of and the number of species and the four different growth preferences of course for high growth preferences then it it's always a all of species stable you can get down to as low as a growth preference of a five and you still have a you know quite a bit of species in a party okay and so so so when lessons here is actually you know unlike the typical consumer resource model if you if you're putting a grocery that's very large you just wipe out everybody else but in this case it doesn't because even you can grow fast you're limited to one zone okay so you do not take over the world right so then we can so so this is a still very simple in that there's no frustration in the system it's just diagonal model right so now we can build in frustration by say basically populating the script with a random plus and minus everywhere okay and there are different ways to do it I'm just puzzled by your statement that your superbug will not take over in a real community it will because it's not following what's a real community well what is a real community is the one where the states are defined by the resource depletion times so you somehow made a transition from resource depletion times to this artificial thresholds and you made a caveat that this artificial thresholds in biomass are an emergent property once a community self-organized to a state it decides on those thresholds which you put a point one point two point three in reality it's the resource depletion times which self-organize we have to have a reality check first what is real what you think is real may not be real I mean consumer resource model may not be real right no no no but all I'm saying is that this is kind of a toy model with this thresholds imposed on it and then you say in your toy model the world it doesn't it doesn't take over right and the underlying philosophical point is that how much of this is a dominate how much of a community growth is dominated by exponential growth that that's the key well you also have exponential growth everywhere in this model it's just that those regions where it grows are defined by those thresholds which are dictated by the total bio you can I can set the gross I mean I can make it no hundredfold what matters in this model is a ratio of the favorable and unfavorable model I still puzzled by this biomass thresholds you said that this is some of artificial you know you don't really believe that in your community which you studied in detail those thresholds are therefore fundamental reasons you said that the system organized with depletion times to give you the behavior you have in a steady state and that I understand and then you kind of switch to this biomass thresholds which is where I kind of lost you yeah yeah no I mean the I mean the so the point and we're trying to understand a community right and and I will say so far there's no understanding we do not understand these things right we have we spent like five years working on one system right and the lesson we got is that it's more complicated than anything the different phases okay and then and then here I'm just trying to try not to describe these different phases but that's the key the different regions of different phases coming from interaction with species and those the then I'm trying to make a simplest way you just using that idea right and let's see what we can get to but the same different phases exist in the consumer resource model you have those times when what we call temporal niches right the the times when a particular subset of new everything we can do here you can do with consumer resource model just much more complicated right so if I want to get that get something to grow in this one region yeah you can play with the parameter the measure get to that right here's just a single parameter then I can use this to explore what will happen it's not possible to explore in a consumer resource model to see this kind of phase no I'm always the favor of a toy model the toy model doesn't explain why the bug which grows fast everywhere doesn't take over that's all I want to say but you can basically you produce some region where I mean if you can cook up a model where where things are cut off right the growth growth can be limited so the Vibrio will not take over in your time yeah the Vibrio will not yeah right so right okay so so so okay so now we're getting so it's not a single parameter model anymore so with some noise right and then so now we look going to do okay so I'm wrapping up just very quickly with two different versions of a random preference model right so here's example so you have basically you randomly put down stuff okay you can imagine doing two ways right one is that the for so each of these states it can support say case species okay then you can look at what happens okay there then then and and then the what you expect is the number of surviving species will be n over k because if case one that everybody survives we're already in this we are just for this priority effect so that every species survive so now you're putting competition and it is doing a bit better than n over k and if you look at who's surviving who's not so the blue guy here the surviving the red ones are the ones that did not survive and you can see very clearly right even though each state in this system every state has a four species of growing right so the ones that survive are the ones that can live in more states so basic generals are selected okay because it's not about competition within but then if I can grow in all of these states well then I get to survive right and then if you take that into account right so then I say each species can grow in case states so they have the same kind of a general generalizability okay they can okay and then you look at it okay so now you can see that the half of the species could survive so as far as many of us a 10 species growing in the same state okay and as you can also analyze a bit to look at what what is a driving preferring the surviving species and that they turn out to be the the surviving species in purple here there well the ones that got extinct they're the one that had to compete with more more species in a given state okay so the average share of a state is the less anyway so the point here yeah is that the you can you can even with competition this diet so basically still gets feature of the diagram although a lot of states are survived and to this towards that end we'll call these community states as dynamical niches I mean I try to avoid the word niches much possible because it's undefined but people have the idea sort of one niche up with species and certainly these states can in this model can many species yeah one one one of these at the gate can support of order one species right okay so I am done the okay you have a question about this about the model yeah okay yeah nice in the sense that so we see many things in experiments even with per wise competition in one resource and for example we see by stability competitive for competitive exclusion coexistence or so we see many things in the experiment so the question is how many of these things this model can reproduce yeah so this model is agnostic to resource there's like the purpose I mean almost so we do not know about resources anyone let's say here you're in a state let's say you have one resource like in alvaro's experiments yeah and for example we see by stability can you reproduce by stability with this model no no so so so in this generalized model that I'm talking about there's no resource so I cannot so it's not intend to explain that you have so many resource that does this and then what happened you put this together and do that right so it's simply it's we're trying to see what we can do without what statement we can make without resource okay and the state the kind of statement I've shown you are the ones that concern stability right how many species kind of accommodate what the interrelation between species growing in the early in the late so forth on how many species can accommodate what about a competition of within a niche that kind of a statement right but absolutely nothing on resource but the growth and outcomes could be due to resource that could be due to you know any number of other things yes but even in this let's say word where we don't care about resources you don't always see coexistence so what when you when you have two species you put them together whatever whatever is the medium and not always see coexistence or competitive exclusion there is a third state so it's like you know for the logical model that is that I already know there's a phenomena right and I want to say something more about well if there's one of the phenomena is going to happen well how does it happen and and you know what happens when you change certain features okay but it cannot address me if you put something together whether whether the system stable not stable no I really assuming they're stable that we're applying a cyclic constraint given that I know the system is going to be stable that could be arbitrary complicated here's a simplest way to produce such a thing right and then what was an interrelation it's kind of a it may or we already have a lots of model trying to answer these special questions we're trying to just look at the other type of a question interest so let me let me finish I'm gonna anyway to discussion I think a lot of a point is about the philosophy underlying the right okay right so just so some of this on my last slide right so I would describe as a phonological model right and it allows us to do dynamics in a simple way right and so the community states right seeing as a dynamic conditions and since it come from the combination of the different phases right so it's it's gonna be if you have n species at least you have the lease of order of 2n of such states biomass density right is a designated state variable means both for mathematical reason and I think for biological reason we're gonna pick one thing it could be that mechanistic if we know cells have a ways to measure the total density super sensing and so forth okay but we're not insisting that's only it's an example of some some feature that reflects the overall state of the community right and yeah so I think what what do we gain with this model I think mainly is a convenient way to explore a relation so as I said anything that's been done here could be done by a consumer resource but but nobody will use consumer resource model to ask this kind of question the you can you can randomly generally parameter right but it's difficult to get into this type of a complex dynamic regime okay the right so and a number of specific things we find from from the specific model we use right one is a quantitative description of the early bird effect and it's a criteria right a the that the the no baby in action the super bug is limited in its its effect and the no the then that the interest state competition and I think it's like that for this random model but all of this is done with very few parameters okay so you can you can make statements so an if I finally I think this was constructed on purpose to shift away from taxonomical classification or I need from from this promise bug is I don't need to know it's the first name last name just how fast is growing in some in some conditions okay and yeah so it can predict community dynamics in such a simple world and it could be rather complex and and but I think the important way it's caused for certain measurements right it's so caused for measurement in between the in between the cycle the different regions and the the grocery and and the the the grocery of each species and the biomass of the species okay and given these in immediate one can make predictions right and about the dynamics and when compared to experiment that's the that's features okay so yeah so Sergei yesterday I talked about the difference between astrology and astronomy right so I think we are still very much in biology in the in the left on the left hand side we care about the names of things the names of the new chair name of the organism and so forth right but hopefully at some point we can transition to a world where we're describing this by mathematics but to do that we need to find some variable right so I'm not so certainly this is a very first attempt but we're trying to do it without using names without using resource and we're trying to just introduce some numbers so this is fun work being done together with Avanish but he just graduated his is a Stanford right now and Arvind Murugan at the Chicago and we have had lots of discussion with the colleagues at the interface of the ecology and physics and I hope to get a lot more input from you from here thank you