15:06:20 All right, go ahead.
15:06:21 Thank you. To organizers for letting me present my work.
15:06:26 So my project, be mostly interested in how photographs. Couple of bacteria are heteros, and if there is any way to control this.
15:06:38 And by the end of this talk. What I'd like to convince us that it's the nutrient availability exogenous nutrient availability, that controls this coupling.
15:06:50 So, just to motivate.
15:06:55 As we've all heard before, microbial folded rows are responsible for fixing almost 50% of the global carbon dioxide, and most of the carbon dioxide that they fix this is at least the header cross which they'll go back as carbon dioxide by respiration
15:07:15 or biomass. And this for the synthesis can just happen on its own, they need the hetero heteros usually provide some vitamins or other nutrients to the photo tropes for them to survive.
15:07:27 And despite this importance. The in lab the studies of photograph hetero communities is not not been that well studied.
15:07:38 So for, for instance, even like some basic questions like, do photos, actually affect microbiome assembly, or bacterial composition around them that's not that doesn't have satisfactory answer.
15:07:51 So if you are a lot of people that say yes and a lot of people that say no.
15:07:57 For instance, for instance, in this study.
15:08:03 They considered three different phytoplankton species. And they, they start analyzed the bacterial composition associated with them.
15:08:11 And what you have this study they say that it is very specific the bacterial community is very specific, because you observed that the the shapes of cluster together and eat shape corresponds to
15:08:26 the different phytoplankton.
15:08:28 However, there are other studies which say the opposite. For example, these guys studied.
15:08:34 Another set of part of plant species and they find that like, if you analyze the bacterial community associated. There is a lot of mixing so it's not like well separate.
15:08:48 So the question that whether there is a effect, and if so, what controls this effect, still open to questions.
15:08:57 And what we are trying, trying to do is answer this question by using a specific lab strain of algae, the Trump climate don't wanna stand. Yeah.
15:09:07 So, in this talk, I might say algae car climate climate of one as they all refer to this one thing.
15:09:13 So what we do is grow soil bacteria in, in association with these algae.
15:09:20 So in particular, we go and collect some soil, bring it back to the lab, this soil will typically have eukaryotes Panda and so on. So we add some drugs to try to get rid of them as much as possible.
15:09:34 And hopefully end up with just header traffic bacteria.
15:09:38 And then we add our lab strain of clammy, and we use that, that culture then to inoculate for different conditions so we start, we have a vial of where we have this climate plus soil and the media.
15:09:55 And we grow them for three days, or six days, nine days, or five days. And after this growth period we do a serial dilution and repeat this whole process 10 times.
15:10:07 So these are what we.
15:10:13 Sorry.
15:10:13 One is the 64 is the factor of.
15:10:20 What is it, it's a it's a minimal Media tab.
15:10:23 It's, it's designed to mimic the freshwater media, so it has it has some trace minerals for the algae and some ammonium chloride phosphate, as phosphate as the phosphate buffer and some goes.
15:10:40 Yes.
15:10:41 And I will talk about glucose shortly.
15:10:46 So this is registered, and we have a LED light panel at the bottom to mimic the day night cycle so it goes on for 12 hours and goes off and
15:11:03 And
15:11:14 Yeah, I will come to that in the next.
15:11:18 since you mentioned glucose is the, the carbon sources for the rest of the population dominated by that glucose or the climate is.
15:11:19 Okay so and then for two different soil samples so be repeat the whole thing for two soil samples. And for each of these growth periods we have control communities that don't have any algae.
15:11:32 And in total we have 60 communities and be secret sequence these every other round, so we have five rounds of sequences for 60 communities each, so all my talk is going to be mostly based on these 16 s.
15:11:49 What are the arranged as a reach.
15:11:52 Oh, good question. They is depending on the growth period so the 12 days reach pretty high od like maybe point 5.6 like that.
15:12:08 But the three days that they don't reach that high and PhD mountain now yeah PhDs, when we start with starting pH of 7.2 but by the time. So for three and six days by the time the end it's still like 6.5 or so, but the 12 days they tend to lower a little
15:12:18 bit like around six.
15:12:21 Just as you keep track of the clammy and the hetero traffic bacteria separately in terms of the ability Are you relying on the 16th. We have to tell you, sort of how much of each Is there a.
15:12:35 So, we don't do it, we don't have a plan we only know I'm asking like so, can you use for example the observance of chlorophyll to detect how much climate is there at the end of the growth, yes yes absolutely so not absorbent we do floors.
15:12:47 Yeah so chlorophyll is the art of philosophies.
15:12:54 Okay, so why do we do these, these different growth periods. So, our hypothesis is that the exogenous nutrient supply is going to control this coupling.
15:13:04 So what this means is that the three day dilution period, we don't be supplying supplying nutrients very quickly. So we don't expect the climate to have a very strong influence on the community, but as a 12 day growth period.
15:13:19 The most of the exogenous nutrients will be consumed and the bacteria will have to depend on the, the carbon fixed by the enemy. So for example, here you see a BOC measurement by a different group, over time, and by 10 days.
15:13:37 It's about $10 million carbon
15:13:41 up, you look at the OD with time curve when, when does it saturate as such that after three days of it for three days. So we don't be only look at audio at the end of each town, so if you don't mind if you haven't looked at the OD overtime
15:13:58 dissolved organic carbon.
15:14:02 So yeah, so we supply very limited organic carbon initially so that's what I mentioned about glucose. Glucose concentration is pretty low, so that we hope that they interact strong.
15:14:20 To me, well and low compared to how much it can produce an end but it's comparable
15:14:31 minimal Adam will okay.
15:14:39 Okay, so.
15:14:42 So next I'm going to dig right into the to the results.
15:14:45 So the first thing we observe is that the algae affect the community composition, but at long timescales. So what I mean by this is if you have the hybrid community which is just the soil back, it's the soil bacteria plus the algae versus the control
15:15:03 community where only the soil bacteria the two communities become very distinct at long timescales.
15:15:11 No same.
15:15:13 So what what I'm showing here is on the y axis is the relative abundance. And on the x axis or the replicates. So, if you consider a three day growth period, you see that the hybrid community and the control community have very similar taxonomic compositions.
15:15:30 And this the same if you go at around 10 they still like more or less look similar, but as if you look at the 12th day communities.
15:15:39 The difference between the hybrid and control is pretty strong by around 10
15:15:48 level. Now this is the otter you will.
15:15:53 Yeah, so basically that's what I mean by long timescales which is a 12 day by around 10 so where they have very limited exogenous nutrients.
15:16:04 So to make this more quantitative, let's let's consider the distribution. So let's take the hybrid communities have given soil type, and given growth period and plot the distribution of the distances so here we are using Jensen Shannon divergence.
15:16:21 And if he also then plot the distribution of say the control and hybrid, if the two distributions overlap. That means that the hybrid communities aren't that different from the control community, or if it looks like this, then we can say that yes the
15:16:38 hybrid communities are definitely distinct from the control communities. So what I'm going to show you next is this difference of the mediums of these distributions.
15:16:49 So the median of the black distribution minus the median median of the industry.
15:16:55 So if that that is more or less zero, it means it's this situation by the two distributions overlap and the two communities are not distinct.
15:17:05 Whereas if it's positive, then it means that the two communities are distinct.
15:17:12 So, what's plotted here on the x axis are the four different periods, and on the y axis is the difference in the medians, and each different panel corresponds to the dog died down.
15:17:25 So at three day you see that for all dilutions down sequenced the meat it's very close to zero, the difference in the medians. But as for the later nine day and 12 day it's much higher, which means that the that the climb is definitely affecting the community
15:17:45 composition, but only when nutrient availability.
15:17:50 To further understand this, we did a PCA analysis of our data, and in light of. Can you give me a sense for what the standard deviations are so I'd like my things dimension less so medians and units of maybe standard deviations like.
15:18:07 So this was done by bootstrapping. So we have the, we have compute the general channel divergences and then we do the bootstrapping with replacement and find differences and so the standard deviation is then the standard deviation of this is the metric
15:18:26 dominated by just a few of the most abundant me she's
15:18:34 usually these Shannon like measures.
15:18:37 They only care about the high frequency, the high of the abundance species, and yet essentially you're measuring changes in a few abundance species or is it telling you something but I haven't looked at that, But we do consider the entire community.
15:18:52 We don't like the quantity that you compute whether it cares about the whole community, it did, yes, it took into account the entire cast. It took into account but does it depend on it.
15:19:04 Okay, if you haven't checked
15:19:20 the control. So the control is not going to be plotted here because control is in this this distribution. So we're finding the, let's say you have like a group of points somewhere here which are the hybrid and the controls are here.
15:19:26 So we're finding the distance between these two. and that's the that's the black distribution.
15:19:34 I see so I should be just looking at the fact that the distance systematically increases above zero. That's right. Okay, good.
15:19:44 Is there a particular reason why you care about the sign of this thing like questions you just take the magnitude of this distance. So the sign if it's positive, it just means that the control hybrid distances, larger than the hybrid.
15:19:58 I mean if it's negative it actually, it means that, so this this distance so they're actually closer to the hybrid than they are to banks themselves. The replicates are far apart from more far apart from each other, compared to the control.
15:20:15 That's like kind of.
15:20:23 Okay, so a beaded pcn this and in light of yesterday's insightful talk I want to mention that we are aware that these are a compositional data so we have taken into account the agents and transforms.
15:20:41 And this all our data is using that.
15:20:45 So here we present only PC One and PC to, because we checked this by plotting the eigenvalue distribution. So if you plug that you see that there's a bunch of stuff last like small values and to eigenvalues far away.
15:20:59 And then if you do a random shuffling of your data these two eigenvalues far away basically going to this.
15:21:06 So, we consider only these two.
15:21:12 The logarithms of the ratios of abundance is that at the center lock transform.
15:21:21 So it's so it's sensitive to things that whole ranges have been, yes.
15:21:28 Okay, so there's a lot going on in the slot.
15:21:41 There's a lot going on in the slot so let me walk you through it. So first of all, the circles and triangles, they represent the two different soil types that boost and the different colors represent the different growth periods, the blues, or three day,
15:21:55 oranges, or six day greens or nine day and the reds and the largest points that you see the largest markers around 10 so what up what the composition is at around 10, and the intermediate markers here there around.
15:22:11 Well that's like, that's the first round that happens, and everything in between is the intermediate rounds. So if you consider these blue triangles here these are round two soil be greedy and they move along the directories and end up here.
15:22:28 And in the same way for example these red, red, red circles here are solid 812 day at two, and they move along these trajectories and end up here by Toronto.
15:22:40 So first thing to note is that at round two. These two soils are very different, very far apart. But by around 10, they converge.
15:22:51 And that's very surprising.
15:22:53 So I mean, starting from such diverse such different compositions, and we do the same thing for the control communities. So keep in mind that the PCA is done on everything at once I'm just plotting them separately.
15:23:08 And you see that for the control communities there is no convergence. So, clearly, whatever is driving this convergence must be the algae. That's just what I understand, this presupposes that there are highly overlapping taxa in the soil to begin with
15:23:28 no there there are two different soil samples and they were not. I should have included they're not there, but the but the fact that they converge means that, that there's something, there's some material with which they can convert right like right meaning
15:23:41 meaning that they have. So you must be aggregating the text at some level, that allows for that to happen. I'm going to talk about that.
15:23:53 Is there a way that while you're doing this experiment on the left you can take the algae out, and then see whether your hybrid communities will sort of start drifting like your control communities, think I'd be positive, in the word, but it's gonna be
15:24:06 hard. And it's also very likely that some bacteria actually are attached to the algae.
15:24:18 So we might end up removing them as well. I see so you can do it like by sighs filtering the energy. Because, I mean we could probably try to remove the algae but probably take away the back
15:24:32 that yes, we can add the drugs and kill him that I may miss but do these algae and soil Coker in in nature, these two, they do. So, it is commonly found this is found in soil, but the the strain that we're using is a lab sprayed, but if they do occur
15:24:52 and okay but these would have not met. Previously, before you combine them.
15:24:57 Maybe not, I'm not sure like be sampled soils from if they were already.
15:25:05 I'm still trying to parse your left blot, again, you yourself admitted that it's very very busy maybe for a paper you should make it slightly more, you know, stratified.
15:25:17 So am I right in saying that if I only care about the sort of the rounds last round of the rip off your serial dilution Site Builder will be basically three groups, one in the lower left which makes us to soil types, no lower left.
15:25:38 This one, which makes us triangles and circles.
15:25:41 And then to others in upper, you know, in, you know, upper part and the lower right, which are unique to a particular soil sample so it's like three possible stable states if you want to have this system, I know so so
15:26:00 so let me just go through it again. So these are the two soil types of circles and triangles. So, initially at round two, you have to look at these small smaller triangles and smaller circles.
15:26:14 So, initially, they're very distinct the two soil types of, I think, and the big the big ones, the big set markers are at around 10. So what I'm saying is, by around 10.
15:26:28 Three days do not converge, the three day growth periods, but the six nine and 12 days converge by.
15:26:37 Okay.
15:26:40 So, all, all 12 and nine days are grouped together in this orange, green, red cloud. Okay, I think I know
15:26:52 exactly why why would they convert maybe they'll never be they might but in in the scale of our experiment they did.
15:27:00 So it looks to me like on both of your plots things moved to the left over time. But if you go back and look at the composition, can you make any sort of statement about what's happening.
15:27:18 Yes, I will, I will come to that. Okay.
15:27:23 Okay so, so, so first thing was that we saw this convergence and soil type so we went back and looked at our composition.
15:27:34 And you see that, although I think this is what the question, some of you asked.
15:27:40 So, so this is again the relative abundance versus the replicates for just the hybrid communities which we know convert. So, if you look at the 12 day for solid a and the 12 day for soil be there's maybe one or two dominant taxa but not that many dominant
15:27:57 taxa that are common. Same for the 90.
15:28:00 So yes, it's a question that, how is this convergence happening. So to answer this, what we did was we be selected some, what we call the convergence tax.
15:28:14 The way we did this is we find the taxa that.
15:28:19 So we first of all we note that the convergence is mostly along PC to.
15:28:24 So we take the eigenvector along PC to whose buildings correspond to the OT us, and refine the otiose that contribute a lot to this distance in at round two, but contribute very little, the distance between the soil types at runtime.
15:28:46 So, so, in principle, in essence, these are the two us that have very different abundances of the transformed abundances at round two, but they become very similar, at around 10.
15:29:00 So will be so we get this convergence taxer for each different growth periods for 12 day communities, you get a set of tags that can that help in this convergence.
15:29:12 For nine day for six day, and we look at how these taxes, change over time.
15:29:19 And what I've plotted here is, this is the list of the convergence tax or that that was selected that way. And all of them, decrease in time, so there's a very strong down regulation.
15:29:31 And so yeah what I'm plotting is the law of the relative abundance.
15:29:36 And if you just as an example I've overlaid a couple of or two random, or to use so circles and triangles are very far apart at round two.
15:29:48 Round 10, they converge. So this was surprising in the sense that they don't up regulate the same taxa to do converge taxonomic Lee but they down regulate the same taxes, the climate basically is kicking out tax that doesn't like.
15:30:04 And that's what's actually causing this convergence.
15:30:08 So if you think of.
15:30:10 So there's a, there's another related point to be made here which is about functional attendance. So, if you. So you saw on the PCA plot that the two soil types converge taxonomic Lee, but it's not converging taxonomic Lee by selecting the same dominant
15:30:27 tax. So the taxes that, so basically have two different sets of tax and the two soil types, which are essentially doing the same function which is eating up the collects it.
15:30:39 So, there is, there is still functional redundancy happening.
15:30:44 Can you go back one slide.
15:30:48 Okay, so I didn't understand here why you focused on PC to it seems to me that in pc one, both of them move. Yes, there is only one of them moves. That's right.
15:31:00 So, so the idea is like if both of them just moved on PC one, they wouldn't convert so it has to be, it has to be movement along PC, I see, I see I see but but just to be just to say more we also did the same analysis along pc one and we get the same,
15:31:13 like, we get a list of tasks and they're all down regulated to the convergent I mean you're probably going to get to this but the convergence. It doesn't have any aspect of overlapping like a positive aspect it only has this negative filter, like ecological
15:31:27 filtering aspect, yeah.
15:31:29 That is nothing that's up regulated it's all done.
15:31:33 Okay, or because I'm confused which side of things you said the climate kicks out the ones that doesn't like.
15:31:55 or is it the others that are happy with whatever the climate is providing.
15:31:56 Too bad. I don't think I understand. So what if you look at the abundance of the the climate is through the round.
15:32:03 How does that go, it's actually more or less what sense is the climate kicking out ones, it doesn't like.
15:32:10 I guess I would have thought that would be the words from my us, if the amount of the climate actually went down, which I was doing better later.
15:32:18 It's some other ones that don't like being with the climate you're dropping out. Oh, I see. I guess that's a question whether the directions that thankfully records both ways.
15:32:34 We don't see an effect on the climate says but we do see the diversity false diversity of the communities which have climate. So the client. So what's the climate is not affected much.
15:32:41 So if you look within the group that sticks around until day, 10 and just look at their relative abundances amongst themselves. Do you, do you still get sort of convergence in that subset.
15:32:54 So you mean, just, Just use the ones that can.
15:32:59 Yeah, so if you go to the next slide right you were saying these convergence tax all drop out, right, so now I'm saying pick all the ones that are still around on day 10.
15:33:10 And then, because it's compositional data you can look at their relative abundances with respect to each other on in previous rounds around 246, right, and you can look within that subset whether you tend to drive convergence or not.
15:33:22 Okay yeah we haven't done that.
15:33:31 Okay. So, so far, I've been talking about convergence and like somebody asked, I'm going to now focus on, they also move right within from round to round they move in time.
15:33:43 So what happens there.
15:33:47 So, for dynamics. What we do is essentially the same. We consider the eigenvector along PC One and PC.
15:34:02 And we be fine the contribution of each other to you to do the movement from say down to to run for around four to six dancing around eight and so on.
15:34:08 And then we pick the otiose which have the maximum contribution, because this motion.
15:34:15 And in this in this condition we have to also do this too for the control communities because they also for the convergence they didn't convert so we didn't have to worry about it but here the control communities also.
15:34:27 So we get us the list of taxes for the control communities and for the, for the hybrid communities. So we pick the tax or that's present in the hybrid but not in the control communities as the ones that are responsible for the motion in presence of these
15:34:47 we call the dynamic stacks.
15:34:54 Do you have a sense for what the participation ratio is and each of these principal components is it like 10 taxa hundred taxa oh so the Yeah, the eigenvector is around.
15:35:04 So, so, yeah, we have like around 200.
15:35:10 So if you, if you find that if you take the dynamics taxa and plot them, their relative abundance, as a function of the dilution rounds you see kind of succession dynamics I'm only showing it for that five day period but this is the same whether you do
15:35:25 it for nine day or 63 days.
15:35:29 So, yeah, this is, this is where we are right now and what we are trying to. So I hypothesis is that the ones that were selected. Later on, which felt the maximum pressure selection pressure would be selected to grow on the algal exit rate.
15:35:46 So our hypothesis is that the ones, the isolates that show up to us that show up here should be better at consuming all the laksa days, whereas the guys that are up here should be generalists they should be able to offer forever can find.
15:36:00 So, what we're doing now is to have isolate so we have the managed part of last year to isolate a few or to use these squiggles are the ones that I have in my lab, and we're trying to.
15:36:15 So what was the other experiments we are doing this to have accidents of algae so just grow algae by itself for 12 days.
15:36:21 Take the spent media and do a respiration acid to see if these isolates how well they grow on these accidents.
15:36:30 And hopefully they will confirm, hopefully they will confirm
15:36:44 the also see the succession dynamics in the control communities or do you not see them okay do.
15:36:49 Okay, so then if you do see them then can you distinguish between, you know, one other hypothesis could be there or you know some of these tax, producing something that the next round of tax or eating which are producing something, the next round.
15:37:01 Can you distinguish between that and the, the fact that these later tasks are actually consuming the Alba likes accidents. Yeah, It's so that's why be.
15:37:11 in the controls, these are the taxes that are not responsible for motion and the control communities.
15:37:26 But, but, yeah, I get your point and I don't know if we can directly answer.
15:37:33 This is sort of related question, wouldn't it be much easier if you didn't have this glucose, that in the beginning, in other words you really are after just which bacteria grow on dissolved are getting carbon exclude you know excluded by by by algae
15:37:50 and heaven this glucose complicate things because it allows for other species to feed and maybe even cross feed yeah it's it's possible to do that but I think we did a control experiment and found that if we didn't have glucose for example in the three
15:38:07 day community there was no od visible only produce. Couldn't measure any change in the after three days so we thought we'll have.
15:38:18 Is it possible that it's because you, there is this time lag. In other words, it takes quite a bit of time for LG to to excrete the carbon so if you just add but microbes a little bit later not you know not not on day one, but on the.
15:38:33 I don't know six. Yes, yes, if you had Wait, wait long enough, it's likely that.
15:38:39 Right. So do you do you think of doing this because it would have been much cleaner, you don't have to worry about it a lot of things you currently worry about.
15:38:49 Yeah.
15:38:56 It doesn't matter to me there's much interaction because do the the the climate can may has everything it needs because you're providing phosphate and nitrogen and everything.
15:39:07 I usually will talk about the collaboration is the answer choices are providing all this stuff, but you're providing the in the medium.
15:39:28 Unless Unless the hangover doing something bad for the climate, climate is good just get everything and do their do their usual thing. And, I mean, the simplest experiment I can think of just to just scroll climbing this medium and take the take the the
15:39:31 spent medium and then see what what what what role is some of what your, what your soil samples. Yeah, yeah, that's fair point and be expect that it might be providing some carbon dioxide.
15:39:42 Maybe not much.
15:39:44 And also.
15:39:47 So, we have, we have the Spanish media experiment, it could be possible that bV have no way of measuring it right now but it could be possible that the span media is different when there is President there are bacteria present.
15:40:00 bv don't know that.
15:40:05 Okay, so, by vivre Viola in the lab work on these experiments with the isolates.
15:40:14 We try to make a toy model, where we have just three species the algae accident eater and a glucose eater. And he, as he basically grows with the logistic wrote the accelerator just.
15:40:30 It's the accidents and the glucose meter justice glucose tax rate or just the accident.
15:40:42 And the accident is secreted by algae at a rate, he and his consumed by the accidentally turned the glucose, just consume.
15:40:48 And, like in the experiments we do see a dilution so all of these, except for glucose everything else is diluted by the same factor. and we repeat the serial dilutions multiple times until we have.
15:41:02 Sorry.
15:41:02 It's on the previous thing but. So, in the hybrid communities at the end.
15:41:08 Is there a way that if you remove the clammy and now passes them and just a media debate, do you expect them to change a lot of will they just sort of stay where they are.
15:41:16 Good question. We don't know we haven't done it was he was, I guess I'm.
15:41:20 Because I guess I'm maybe relatively naive question but if you take the climate and grow them up on their own, and in some way, he killed them such that you don't break the complex molecules but you just provide like dead clammy like shells of clammy
15:41:34 is that more useful in terms of what these bacteria that is selected are going on in terms of yeah broken starch and things like that. The we are not sure that Cameron wants to excrete pompon So, and that clammy, it's hard to break the cell walls right
15:41:52 not not everybody has the capability. So I think it will be different if you just provided that to me was, because you said you kept track of the Clemmie od like is the relative total abundance of plumbing roughly equal to the total bacterial biomass
15:42:07 versus like is the bacteria way higher than the plumbing, that's, that's difficult to say with the data we have this one is Laura sense then others od and when you measure od of these communities that claim is still inside, so it's hard to distinguish
15:42:28 OE sure from zero to 12 days it changes but I was I meant within with between rounds of dilution it doesn't change.
15:42:40 Okay, so with this model BB do serial dilutions glucose goes back to its original value just like doing the experiments. And if you plot the steady state abundance we kind of see what we expect that three day growth paid the extra data has very low abundance
15:42:55 that has had the 12 day growth period, it increases in abundance. So, kind of like the succession dynamics we see the x, or we hypothesize that they accidentally selected for the later.
15:43:07 So with this I just want to conclude that we have hope I've convinced you that the nutrient availability controls the coupling of bacteria and the and folate growth, and it affects the community composition.
15:43:23 It causes convergence of divergent of different soil types, and we observe succession dynamics, and hopefully we understand these with the experiment and I split like to thank everybody love especially suppose a fantastic advisor and all members of the
15:43:41 lab for their support and corporations and thank you all for your attention.