Skip to main content
Loading Explore guides…
Climate Change Impact on Lakes · Acedra
Home
Explore
Biology
Climate Change Impact on Lakes
Sign in to save
Chat
If YouTube asks you to sign in, open the video on YouTube.
Watch on YouTube
Reload player
Transcript
160 segments
Copy
Hide video
0:01
Zoe Rand: Okay. Hi! Everyone Welcome to quantitative seminar. I'm Zoe. I'm. An organizer this year, and i'm going to get a started with a land acknowledgment. The University of Washington acknowledges the coast sales peoples of this land, the land which touches the shared waters of all bands and tribes, and the suicamish to lay up and knuckle, shoot nations, and them also acknowledges that we live and work on the ancestor lands of the Duwamish people. The first people of Seattle.
0:29
Zoe Rand: Now, i'm really excited to introduce our speaker today. Celia Simons is an assistant professor at the University of California Irvine. Her research focuses on the effects of climate change and other human impacts on lakes with a recent emphasis on how global change drivers
0:44
Zoe Rand: alter species interactions. She did her undergraduate and master's degrees at Queen's University, under Dr. Shelley Arnut Supervision, where she researched the functioning of tundra pawns and their response to salinization.
0:57
Zoe Rand: She then earned her Phd. At Uc San Diego, working with Dr. Jonathan Sherin, Wh. Where she started studying lakes and ponds of the Syria, Nevada, before she started at Uc. Irvine. Celia was the University of California, President President's Post doctoral Fellow at Uc. Santa Cruz.
1:14
Zoe Rand: She has continued working in the in the Sierra Nevada, combining observational and experimental work with synthesis projects to scale up to larger spatio, temporal scales. Celia would like people to hold their questions until the end, and so when that time comes online folks feel free to submit to the Q. A. And in person folks feel free to come up here to ask your question and thank you so much, Celia. Take it away.
1:38
Celia Symons: all right. Well, thanks so much for the introduction, and thanks everyone for coming to the talk. I'm excited to be here virtually with you and talk about some of my recent projects.
1:50
Celia Symons: I will be presenting some work, looking at how species, interactions and population stability vary with the environment.
1:59
Celia Symons: So before I get into the talk I thought i'd start with some acknowledgments and a bit of a history on how this project came to be so both because I want to highlight the people who are involved, but also because this program, Eco days. Celia Symons: is an opportunity that's available for people who are about to finish their Phd. In a quant Sciences.
2:19
Celia Symons: or people who have just finished their Phd. So if there's anyone in the audience who's in that category, I highly recommend this opportunity. Celia Symons: This is a program run funded by Nsf. And as well, and run out of the University of Hawaii. and you could say, what you do is you propose a project that you're interested in doing so? Some sort of synthesis or a conceptual piece typically that you're interested in?
2:43
Celia Symons: Then you get to go to Hawaii with about 30 other early career scientists. And you picture project. Celia Symons: You join each other's groups, and you write collaborate papers. So all of the work that i'm going to present today actually began at etern days in 2,018,
2:59
Celia Symons: and the project that I pitched Celia Symons: for people to join at e for days was a project looking at the top-down and bottom-up effects of so for linkedin and cytoplanes. I'm a freshwater ecologist i'm interested in these dynamics generally, and my project that I pitch was looking at how Celia Symons: interactions change through time and with the environment. Celia Symons: So I wrote a proposal using some kind of basic ways of looking at time varying interactions.
3:32
Celia Symons: And I made this pitch. And so the kind of data I had, which was the time series of plankton and a number of lakes throughout the globe. Celia Symons: and immediately after this I had 2 people very interested in the project. Celia Symons: and those 2 people were Tony Rodgers, who at the time was working with Steve Munch and I'll create a here on who at the time was working with George.
3:54
Celia Symons: So if you know those 2 names. You might guess that these 2 people came down very excited to talk to me about a period dynamic modeling. Celia Symons: a technique that they were both using in their postdoc and Phd. Work.
4:07
Celia Symons: So, although it was a different quantitative framework that I proposed, it was kind of the perfect trading of forces, and through this chance meeting that allowed us to do all the work that are going to present today.
4:21
Celia Symons: The 2 main collaborators in this work are Tony Rogers, who is now a research apologist at the Noaa Celia Symons: Noah. Sorry. So it was fishery, science on it. and as you, Munch, who is also at the same location as a program way for the system. Forecasting
4:39
Celia Symons: this works is also based on publicly available data, long term monitoring data that has been collected by many people painstakingly over many decades. And so I also want to acknowledge the huge contribution of those data sets and our availability to this kind of work, and i'll be noting Celia Symons: so specifically throughout the talk.
5:05
Celia Symons: so jumping into it. Part of my talk today will be addressing one of the oldest questions in ecology. Celia Symons: What controls the distribution of biomass within and among ecosystems. Celia Symons: There's some areas like the forest on your top left where there are a ton of plant life. and others like the desert, where they are more to pop from. There's also a lot of variation in the relative biomass at different levels.
5:31
Celia Symons: So, for example, famously. The African Savannah support a very large biomass of animals relative to the plants Celia Symons: compared to places like the temporary forest. Celia Symons: so I'm. Generally interested in determining what causes this natural variation.
5:49
Celia Symons: So let's zoom into a specific example. So here I have a Help Forest, where we can see the kelp and fishes. Celia Symons: But there's all the other marine life that you cannot see in that picture. So the question is, what determines their abundances.
6:08
Celia Symons: but the most basic level, the resources that are available in the ecosystem can determine the biomass through bottom up control resource, control. Celia Symons: She'll be using interchangeably. Here we're the amount of primary productivity determines how much biomass can occur in
6:31
Celia Symons: on the other side. There's also consumers or predators in these systems that can also control biomass through cascading top down or consumer flex Celia Symons: through their prey all the way down to primary producers.
6:48
Celia Symons: The balance of these, top down and bottom up pathways in food web are important to understand the and the because the relative importance of these is likely dependent on the environment. Celia Symons: So as humans shift environmental conditions. Celia Symons: Not only are we directly influencing the survival of the species. But we're also altering the strength of species in our options which could reshape ecosystems even in the absence of altered species. Composition.
7:20
Celia Symons: There's a ton of strong evidence that both resources and consumers can have tremendous impact on ecosystems. Celia Symons: And as a freshwater ecologist here, I've included 2 iconic pictures from whole like experiments. Celia Symons: So in these experiments the lake is divided in half within an Celia Symons: and in the top example, the you can see the effect of top down control Celia Symons: from a trip of cascade. From As and Passover removed, there is an increase in plankton
7:51
Celia Symons: which decreases the abundance of so function, and therefore it increases the Celia Symons: concentration of all 2.
7:59
Celia Symons: So you don't need statistics to see that this this shift causes a really large rearrangement of the ecosystem.
8:08
Celia Symons: So similarly, there's strong evidence that bottom up control can also really have large impact on links, as shown by the addition of phosphorus resulting in alcohol blooms on the bottom left.
8:23
Celia Symons: Now, although these are iconic examples of how important consumers and resources can be the the story Celia Symons: that they tell Celia Symons: kind of misses, Some other experiments where similar manipulations have been done where the effects are much weaker, or sometimes even have no effect at all. Celia Symons: So how can I be? So? There's a lot of different hypotheses.
8:45
Celia Symons: But one explanation is that the environmental conditions and these systems determine the strength of these species interactions.
8:57
Celia Symons: So zooming them on our separate system of zoom like tennis consumers and fight a resources Celia Symons: the bottom up and top-down interactions between these groups might be variable over space and time. Celia Symons: So these interaction strengths between the 2 can vary with the environment. Celia Symons: One prediction is that hop down control will increase with warming Celia Symons: climate warming because metabolism is more time to care sensitive than photosynthesis.
9:26
Celia Symons: The other thing that might influence the relative importance of these interactions or relative strength of these interactions. Celia Symons: A species trees. So, for example.
9:37
Celia Symons: Vital Clinton can grow spines or grow commonly in ways that make it harder for his opening to to grace on them. Celia Symons: so that can influence the Celia Symons: in fact, they have ons open, so this bottom up effect. and so Plankton would then have a week or a top of that on the as well. In a similar way species diversity in the whole community. Celia Symons: It could influence a interaction strength Celia Symons: largely be a complimentary.
10:07
Celia Symons: So from these experiments there are lots of insights about how the environment influences terrific interaction. Str. Celia Symons: But these experiments are often very limited in spatial, temporal. Still. Celia Symons: the manipulations are rather extreme. Celia Symons: So, for example, here they are removing an entire trophic level, right?
10:27
Celia Symons: And so it's more likely in food reps that they're going to be affected by variation and abundance through time, not the presence from absence of a whole specific level. Celia Symons: The other thing is that experiments aren't always tractable If we're talking about Celia Symons: really large ecosystems or ecosystems where maybe there's some export of considerations about if we can do such an experimentation.
10:49
Celia Symons: So one of our goals is to make better use of observational data to gain insight about species interactions. Celia Symons: Monitoring data is collected over a much larger areas. Longer Time periods catch the kind of the messiness of of nature.
11:07
Celia Symons: and increasingly they are available for us to right. So there's things like the Ltd. Celia Symons: How coffee all these time series data that are increasingly publicly available.
11:22
Celia Symons: So, coming back to this initial setup. We wanted to see how the environment influences consumer and resource control so that we can predict something about how food data structure might change into the future.
11:36
Celia Symons: Okay. Celia Symons: So in the past, one way to look at the relative importance of resource and consumer control in an ecosystem. Celia Symons: it was simply a regression between the resource and consumer abundance. Celia Symons: If they're positively correlated. This is assumed to be indicative of resource control, as when you have more resources, you can have more consumers. So a bottom up effect of those resources on the consumers.
12:02
Celia Symons: And conversely, when you have a negative relationship, this would indicate consumer control, where more consumers result in a reduction of the lower terrific level Celia Symons: abundance so topped out of that
12:16
Celia Symons: one example of this kind of analysis that is really common in the literature. is a paper here in 2,015, looking at the sign of correlations between Celia Symons: many different resource and consumer pairs and marine ecosystems.
12:33
Celia Symons: These relationships vary from positive slopes, so resource control in red Celia Symons: to negative slopes or consumer control and blue.
12:43
Celia Symons: as you can see, there is a ton of spatial variation in these relationships. Celia Symons: And in North America at least, it looks like there's more resource control and lower latitudes and more consumer control of hire a lot of food, so it's totally something about what factors are altering the biomass of these ecosystems.
13:04
Celia Symons: So this is a a useful place to start. Celia Symons: But as you can imagine, each location Celia Symons: is only giving you one correlation value. Celia Symons: so it doesn't tell you anything about temporal variation. It doesn't tell you about the relative importance of resources to our control. And certainly not if they're changing over time. Celia Symons: which is what we were interested in.
13:30
Celia Symons: So, as I said before the approach that we took to accomplish these goals is to use a method called for dynamic modeling, or Edm.
13:42
Celia Symons: And Edm is on the far end of the continuum of models that range from highly mechanistic models where the relationship between each constituent part. So here are species Celia Symons: is known and included explicitly in the model. But the other end of that continuum is these equation 3 methods of which Edm. Is a part.
14:04
Celia Symons: and so historically, with complex systems, it was thought that we needed these mechanistic understandings, and we needed to painstakingly build it up, bottom up by measuring the interactions of each component, each species. Celia Symons: And if we're interested in environmental context dependence. Not only do we need to know something about their interactions, but we need to know how to vary with the environment. Celia Symons: So these easy purchase are are hard
14:32
Celia Symons: to nearly impossible to scale up to natural ecosystems with hundreds of species. Celia Symons: And so, I think increasingly, there is a movement to these equation free methods where we can use time, series data and the information contained in these time series Celia Symons: to understand how complex systems like. If we behave Celia Symons: and predict how they'll behave in the future.
15:00
Celia Symons: So what can I do that we can't do with the correlative approach that I talked about in the beginning.
15:06
Celia Symons: So here's an example with data on resource, abundance, and consumer growth rate on the wide access.
15:14
Celia Symons: So you might think well appropriate, as the consumer increases with resource. So this looks like a bottom up effect Celia Symons: with a lot of noise. So resource control with a lot of.
15:27
Celia Symons: But when you actually connect the points through time you can see it's a little more complicated than that. Celia Symons: So this is actually a complex predator. Prey cycle that comes from a deterministic food chain model.
15:41
Celia Symons: And what you can see is that the relationship between resources and consumer growth Celia Symons: could be positive, negative, or 0, depending on where you are and what you can get where you're on. Celia Symons: And this
15:56
Celia Symons: what Edm can do is then quantify the relationship at each time point. and it does that, using a locally weighted linear regression or other non parametric method. Celia Symons: where the local slow coefficients tell you something about the interaction strength between resources and consumers.
16:14
Celia Symons: So importantly, these slopes are determined, not just by what has happened most recently in time, plus random noise, like some models do. but by trajectories that are all the other times when abundances were similar. Celia Symons: So, for example, in this region, where the slope is negative. we're using information from all the other times where the resource and London and consumer growth were in this weekend
16:40
Celia Symons: we can result cross trajectories by including axes for additional variables or timelines of existing variables in the model.
16:53
Celia Symons: So again, these methods let us look through time and see periods where we have resource or consumer control, instead of just quantifying one metric for an entire time series. Celia Symons: So, using these methods, the purpose of our study the first day i'll talk about was to determine how bottom up and top down on interactions vary through time. Celia Symons: If there are any environmental drivers associated with this variation.
17:19
Celia Symons: and if those relevant drivers differ from our looking across a bunch of really different ways, or how they relate to variation with in a link through time.
17:35
Celia Symons: So, as I mentioned the the focal ecosystem that we're going to be using our Laki assistance, where we have Celia Symons: so Clinton, and fight a function, abundance and environmental variables measured monthly Celia Symons: for at least 6 years. This is again where I want to mention the importance of these long term sampling programs. and certainly these kinds of analyses would not be possible without the data being shared widely.
17:59
Celia Symons: The data mostly come from the North Temperate Lakes, L. T. Er. In Wisconsin. at the St. John's River Water Management District in Florida. Celia Symons: and then individual lake sampling programs in Europe, Japan, and New Zealand.
18:14
Celia Symons: We chose these sites because em does not deal with a lot of missing data. Very well, and all of these sites have sampling that goes throughout the winter Celia Symons: at monthly timescales, and it has been collected for many, many years. Celia Symons: So those were important considerations when we were trying to find data Celia Symons: same also especially appreciative that these spikes are sample throughout the winter seasons as well.
18:47
Celia Symons: So we collated data from 12 to 12 different lakes around the world. And these lakes span a wide range of meteor and conditions. So this plot here Celia Symons: shows the terrific state index, which is basically a measure of greenness. Celia Symons: It combines chlorophyll, nitrogen, and phosphorus into this index. Celia Symons: and tells us whether a lake is a like a tropic, like like how far New Zealand Celia Symons: that is quite clearly blue. From Google Maps
19:18
Celia Symons: up to you trophic like the lakes in Florida. We have, like both cleric picture there, which is quite agreed.
19:31
Celia Symons: The interactions that we're interested in. We're the bottom of the effects of Celia Symons: and the top down effects out. Are Celia Symons: we use clerical as a proxy for final plankton biomass, and then we fit 2 different models for every single lake. Celia Symons: using a locally weighted regression
19:56
Celia Symons: in the first model, the bottom model. Celia Symons: The response variable. Was so if I can growth right. So the amount that the Zoo planked in abundance had changed from the last time. Point
20:08
Celia Symons: your predictor variables for the Zealand and abundance timeline so Celia Symons: so like to abundance. One month ago. Celia Symons: 2 months ago, representing the with in tropic level effects.
20:20
Celia Symons: we then have the core delay abundance one and 2 months ago. Celia Symons: and then 2 seasonal predictors so assigned to little functions to account for seasonality.
20:31
Celia Symons: The cross for the bubble effect of Zo claims or sorry Core fell one month ago on the length and growth rate Celia Symons: is the interaction strength that we're going to call the bottom up effect that we're interested in, and that I'm going to be focusing on for the results. Section
20:50
in the second model. Celia Symons: The predictor variables were the same. but the response variable was chlorophyll. Celia Symons: So I won't go over this modeling as much detail because it's basically the same. But here we're interested in the effect of so plankton abundance one month ago Celia Symons: on clerical birthday. Celia Symons: So those quantifies the top down effect of
21:19
Celia Symons: Okay. So what Celia Symons: what do you get out of analysis like this? So Here's an example from Lake Zurich, and these are the data that we're putting into these models. Celia Symons: We have the total so points in abundance for every time stuff with a little bit of missing data in the middle. and then the a concentration Time series as well.
21:43
Celia Symons: We can extract the top down and bottom up coefficients. So again, those one month leg effects from the other turfic level. Celia Symons: So you get a measure of the interaction stream for the bottom up and talk about effects.
21:59
Celia Symons: so as you might expect, the bottom up effects tend to be positive. Celia Symons: So that means, as you increase the amount of fight of plankton. There's an increase in the growth rate of Z plankton in a classical resource control scenario Celia Symons: and the top down effect Celia Symons: is generally negative. We're more zoom linkedin results in a lower growth rate of.
22:24
Celia Symons: But we also see that there is a lot of variability in the strength of these effects. Celia Symons: and you might also notice that they seem to mirror each other. So that is not something that had to happen mathematically, but just emerged from these 2 separate models.
22:45
Celia Symons: so to see which drivers were associated with variation and interaction strength. We pulled the results across lakes, and modeled the extent to which variation in both kinds of interaction, frank Celia Symons: or explained by seasonality. Celia Symons: the tropical state, index temperature, and still think the species richness. Celia Symons: 3 things that we have hypotheses from the literature about how they might influence this interaction.
23:17
Celia Symons: So we did this first for interaction strengths across the full range of environmental variation across all lakes. Then we're also interested in the within lake variation. And so here we centered Celia Symons: the predictors that included the random effect of like. Celia Symons: and include these illustrations, because they show how these across and within like responses could potentially be different.
23:51
Celia Symons: So what actually explains duration and interaction strength Here I've shown a plot of partial squared or other squared values, indicating the proportionate variation and interaction strength explained by each predictor. Celia Symons: So across all the lakes the trophic State index explains the most variation in both top down and bottom up interaction. Celia Symons: So the whether the lake was the super, clear, blue, or really green green. Celia Symons: impacted. Whether we saw
24:22
Celia Symons: the the type of interaction since we saw. On the other hand. Celia Symons: within lakes none of our predictors explain very much variation and interaction strength between them. But the predictor that explain the most was just seasonality. Celia Symons: and in both cases there really wasn't a strong effect of temperature, as we had expected, or species richness
24:48
Celia Symons: in terms of what those relationships look like here. I part of the mean and standard deviation of interactions strike for each plate across the Turkish State index. Celia Symons: So, starting with the top down effects, you can see that the top down effect of zo plankton on phytoplankton in our most oligotrophic history of poor lakes is actually positive.
25:09
Celia Symons: So that means when you have more. So you end up increasing the growth rate of final content. Does it seem kind of counterintuitive at first? But actually it is consistent with literature of Celia Symons: as a point of nutrient recycling and excretion being an important determinant of productivity. they need super nutrient core systems.
25:32
Celia Symons: As Tsi increases into these visa traffic lakes where most of the previous research on the topic has been done. Celia Symons: we see the more typical negative impact of the so top down effects that we would expect to see. Oh. as more as it's like to know they're They're reducing the birth rate out of front of them. Celia Symons: And finally, in the mostly terrific systems. We see those interactions coming back down to 0.
25:59
Celia Symons: And so this might be explained by a greater proportion of inevitable, better plankton and neutropic systems. A lot of them are colonel. Celia Symons: or or even create toxins. Or it might just be that there's a plankton are not able to insert control over Grant rates at these at the concentrations.
26:21
Celia Symons: Now look at the the bottom up effects. We again see that they mirror each other with the strongest form of effect. It's occurring at me to high Celia Symons: a terrific state index. The effective photo points in abundance on low point and growth rate Celia Symons: is weakest in the low, nutrient systems as well. Much Generally most of these values are positive, indicating that more. But if I can increases the profile.
26:50
Celia Symons: Okay, so it's been appearing across this really wide range of conditions between all those various lakes. But we're also interested in the within like creation. And again, the most important predictor was just seasonality. Celia Symons: and I pulled out 3 examples of like that have this really clear seasonality in interaction.
27:14
Celia Symons: and we see that the magnitude of both Celia Symons: bottom up and top down effects tend to be greatest during the warmest month, so Celia Symons: sometime around May to August, in the northern hemisphere, and in what we would think of as winter months in New Zealand.
27:32
Celia Symons: This variation is consistent with Celia Symons: a seasonal reset hypothesis that can hurt, for which occurs due to successful processes. So i'll describe that a little bit more.
27:45
Celia Symons: So this is on your left an image of the like an ecology group model that you Celia Symons: probably have most recently seen a textbook up. So i'm sorry. And this is kind of the classic view of successful dynamics in Lake E. The systems on an annual time scale. Celia Symons: Here we have the fighter Clinton, and still plankton abundances in the blue and red lines.
28:11
Celia Symons: and the shading under the blue line for quite a pointing represents the palatability or and ability of those fight to plankton with more edible phytoplankton being these single cell, unprotected species, and the inevitable side of painted having more spines or being more funnel.
28:32
Celia Symons: So, as the Celia Symons: ice mount in the systems, we have this rapid growth of edible phytoplankton, at first Celia Symons: followed by an increase in the Zoplankton abundance. Celia Symons: So at the same time, in our model, we see this increasing bottom up effect of Celia Symons: plankton on so plankton Celia Symons: and in large grazers are what is appearing first, and so we see that the top down effect Celia Symons: of these are also increasing at the same time.
29:04
Celia Symons: because of this gracing pressure, there is a shift to more inevitable side of plankton. which ultimately reduces the top down and bottom up effect simultaneously. Celia Symons: So show you all this to just indicate that these species. Interaction strengths that we're getting from
29:22
Celia Symons: just time series of abundances zoom app on to known dynamics and late systems. So that gives us a little bit more confidence in these methods and their ability to recapitulate species interactions that are well known.
29:38
Celia Symons: So to conclude this first part, so we see that species interactions for Celia Symons: terrific level. Interactions are not static in time. and that different drivers are associated with this variation depending on whether we're looking across or within lakes. And so that's likely because of the greater range of environmental variation that's captured when looking across flakes.
30:01
Celia Symons: So when we're going from that hyper related to the system, that's a lot of variation in Tsi, whereas across the year, within a lake Celia Symons: you would really see such a a dramatic change.
30:15
Celia Symons: We also find some results that are consistent with some existing hypotheses, like the seasonal resets, and you through cycling. Celia Symons: but not others. So we really expected to find a relationship with temperature. Celia Symons: But there was no evidence about.
30:33
Celia Symons: And so the most interesting thing is that these findings emerge entirely from observational data. Celia Symons: So of course, we need theory experiments to understand why we see the relationships that we do. But these methods are able to quantify the relationships and test hypotheses within fully Celia Symons: ecosystem. That would be Celia Symons: impossible, or Celia Symons: at least difficult to do experimentally.
31:00
Celia Symons: So we've proposed that these methods and Celia Symons: building on the people who have also proposed these methods, her promising ways to explore natural variabilities. These interactions, and what's great about these approaches is that they are applicable to any kind of species interaction. Celia Symons: not just consumer resources. Celia Symons: So, For example, let's say you're interested in that pollinator interaction.
31:25
Celia Symons: You could use time series data to look at the change in interaction strength between the 2 groups. So perhaps the effect upon a Celia Symons: pollinators is increasing or decreasing over time. That's something you might be able to pass out from time series. So again, this is applicable to a lot of different systems. Celia Symons: and can be applied when you have sufficient data across the relevant time scale for the interaction you're interested in.
31:52
Celia Symons: So now i'm going to switch to a second project. Celia Symons: using very similar methods and most of the same data. Celia Symons: But instead of looking at context, dependent species interactions. We're going to look at the stability and predictability of population dynamics.
32:17
Celia Symons: So the contact context for our interest and stability Celia Symons: comes from calls for ecology to be a more predictive science. Celia Symons: And so to be a more predictive science, we also need to determine what our limits of prediction are Celia Symons: so determining ecological forecast horizons. Celia Symons: I wanna highlight Tonya Rogers, who is
32:40
Celia Symons: a co-oper on all of this work, has a really great recent paper. That analysis, using on the method that i'm going to talk about today Celia Symons: to determine the prevalence of chaos and the Time series, showing that chaos is common in nature. Celia Symons: Sorry I had a recommend talking that on it.
32:58
Celia Symons: So for this part of the project we are interested in time during stability or intermittent periods of chaotic dynamics and response to more stable dynamics. So these intermittent periods one single dynamics would present barriers beyond which precise forecasting possible. Celia Symons: So you're also interested in determining, if we can figure out when to expect stable or unstable dynamics. Celia Symons: and if it was related to variation, the environment.
33:27
Celia Symons: or there are certain times a year, and dynamics are stable, and others when they're on disable? Or is there a certain scale, or which they're stable? All of these kinds of questions.
33:37
Celia Symons: So what I mean most ability. So in this cartoon you have 2 time points in black dots. This is from a time series of abundances. So these are 2 abundances connected to the time. Celia Symons: and a few Celia Symons: Dv. 8 from that exact point. So now you're starting close to that original point but a bit slightly. Celia Symons: But if you have a Celia Symons: stable system possible system. All those trajectories are going to end up nearby each other at the next time.
34:09
Celia Symons: So we're saying, If you have a Celia Symons: chaotic system on a single system. Slight deviations from the starting point are going to amplify. So they're going to result in really large differences at the next, some point Celia Symons: which was obviously make predictions in the system really difficult.
34:29
Celia Symons: So the data that we use for this part of the project is the same data or the same data sour as our last part of the project. Celia Symons: But here we're pulling out just individual species abundances. So the time series of one species from each lake system. We actually also included some ring systems. This time.
34:52
Celia Symons: We're also interested in how the stability changed as you looked at higher and higher terrific resolutions. So we then some together the species abundances Celia Symons: into functional group relevances. So here i'm sharing an example where we take Celia Symons: one coca-cola species Celia Symons: for the species abundance, and we sum together all the couple of pods in the community to look at functional group abundance. and finally some together.
35:18
Celia Symons: all the plankton into a terrific level abundance.
35:22
Celia Symons: We then model these, using that same S. Map, Edm. Approach where we have the time series of the plankton, and we are modeling it as a function of time lines of itself. So the delay question. Celia Symons: after selecting the optimal embedding, come in Celia Symons: embedding dimension and time delay based on ours. we went out consultation.
35:48
Celia Symons: Okay. So what what values are we interested in looking at? And how do we compute that? So Celia Symons: the first thing we looked at was a measure of global stability across the entire time series. Celia Symons: So this is a way to get a metric or an index of chaos. So we computed the we have an exponent or le, as i'll be referring to it. And positive allies indicate long-term sensitivity to the initial conditions. Celia Symons: So therefore they're an index of pay off
36:22
Celia Symons: the equations written out here, and those computed for the length of the Time series doesn't get the Jacobians for each time Step here.
36:33
Celia Symons: And in addition to that global metric of chaos index, the payoffs. We are also looked at the time bearing stability. Celia Symons: So again, we have Jacobians for each time point. and from those we calculate an eigenvalue that is a metric of stability at each time point. So the combination of these 2 stability metrics tell us both the global instability. So it's the long run growth of small perturbations.
36:57
Celia Symons: And then this one tells us the time very local instability for the short run growth, the small at a particular point in time.
37:11
Celia Symons: So here's an example of the type of results we have. So first, we calculate the global
37:19
Celia Symons: stability for one population. So in this example I'm. Showing data from a marine copopod and Port Aaron Bay. Celia Symons: First, we calculate the Le for the entire Time series and find out the value is positive. So this indicates that this is a stable system. It's pretty close to 0, but it's Celia Symons: okay.
37:42
Celia Symons: But Celia Symons: and we look at the time varying instability measures. We find that there are periods of instability in the spring Celia Symons: late late winter to early spring. So you want to X-axis here. Celia Symons: and that switches over to more stable dynamics in the late summer and into fall and early winter.
38:07
Celia Symons: So, using these methods, we're able to document a a widespread occurrence of this intermediate chaos in natural systems. Celia Symons: where we're alternating between periods of stability and instability, and these for really occurring on annual time scale. So you can see Each of these lines across the month is a different year, and there's a a relatively Celia Symons: consistent pattern between years.
38:33
Celia Symons: So we we found out our measures of stability do correspond with predictability in the system and the interpretation of something that is globally stable. But in a minute Celia Symons: I have these periods of Internet in chaos is a little bit hard to think about in terms of predictability.
38:49
Celia Symons: So on one hand, we're claiming that instability leads to a lack of predictability on scale months because of this Internet and chaos. But on the other hand, we're suggesting that the longer-term predictions are possible because this is the globally stable population. Celia Symons: So what what I suggest is the the safest way to think about this is that Celia Symons: we'll probably be able to make more accurate predictions for winter and fall
39:13
Celia Symons: rather than the spring, when things are less stable. Celia Symons: And also that course screen predictions on the annual time scale are feasible. Celia Symons: and that's because of season long term in. So, for example, abundance will go up in spring and down and fall so exactly by how much it might be harder to say.
39:36
Celia Symons: Again, this analysis was motivated in part by environmental change, so as the environment ships. But we see increase or decrease stability of population dynamics. Celia Symons: and going into it, we had these 2 hypotheses, so simulations have found that chaos is most prevalent Celia Symons: at intermediate productivity. Celia Symons: and also that community stability is known to decrease the temperature.
40:06
Celia Symons: So it'll be fine. Well, we didn't find very strong relationship with the environment. Celia Symons: And of course we had very little evidence that stability. Celia Symons: very good product could be at all. Celia Symons: But there were higher, on average, at higher latitudes where there was lower, mean temperatures and more temperature variability. So this is actually the opposite of what we Celia Symons: i'm sorry, disagree with what we might. But again, these these
40:37
relationships are weak and Celia Symons: mostly found. Once you aggregate to the functional group or trophic group level.
40:48
Celia Symons: The more consistent result that we found was that our taxonomic resolution really changed, whether we found the cost or not, chaotic relationships. Celia Symons: And so, if we look at the species Level Celia Symons: Time series, more than half of those were characterized as chaotic. So this is
41:10
Celia Symons: comparing the L. So across the entire time series. Celia Symons: whereas when we irrigate them into functional groups, and even trophic levels, the proportion that we're classified as chaotic, really reduced, especially at the trophic level. Celia Symons: So this is an interesting result. We know that as you increase the tax amount of resolution, you might expect a decrease in the overall variability in biomass.
41:36
Celia Symons: as you are just coming up more and more time series. But that doesn't necessarily mean that we had the prediction that one was going to be more of less stable. And so Celia Symons: I hear this. What this is suggesting is that it's going to be easier to predict tropic level Celia Symons: variables rather than the individual species opposition.
41:55
Celia Symons: And there's some. There's some papers that I found that in nature as well. So, for example, it's been easier to reflect. The timing and magnitude of the spring bloom, title, point in plume, rather than which species actually make up the composition of that room.
42:14
Celia Symons: So from this part of the project we've shown that the prediction, accuracy. and sensitivity change, and therefore potentially management, effectiveness might be greater at certain times of the year. Celia Symons: so specifically in the spring, when dynamics are more unstable. That might be a period where prediction is less accurate and the outcome of management interval functions might be more variable because of the sense of the activity of the trajectories to slight deviations.
42:42
Celia Symons: When one thing on this group talked about a little bit, I think we're pretty far away from, but pretty interesting Celia Symons: is, if we can reconstruct the dynamics enough to actually take advantage of this and stability where we can make the small perturbation, small perturbation in terms of management which would make a really large change the ecosystem. Celia Symons: They're kind of harnessing the power of that instability for our use.
43:05
Celia Symons: The other thing that we show is that prediction might be more feasible for tax economic aggregate, some for an individual species. Celia Symons: and that that's useful. I would say a lot of work already does this. So, for example, in La, because I think there is more work Celia Symons: predicting by
43:26
Celia Symons: another tool that this group of collaborators has developed Celia Symons: something that another thing that can help increase the predictability of the systems is a tool called empty Edm. Celia Symons: Or a. Celia Symons: and this is in a paper that was just published a couple weeks, or maybe months ago. Now.
43:50
Celia Symons: where we use the same Edm methods. Celia Symons: But instead of using regularly space time lags from calendar time Celia Symons: we instead use something. We're calling the metabolic time step. Are we basically rescale time and use shorter legs when temperatures are warm Celia Symons: And I know it's happening faster and longer. Legs when temperatures are cooler and dynamics are happening slowly. Celia Symons: So this method.
44:16
Celia Symons: you know, increase the predictability of almost all of our time series. But it does assume that the the speed of dynamics change, but not necessarily the form of the dynamics themselves. So we'll go further into that on that project, but I thought I would highlight it in case someone is interested in exploring that method. But
44:35
Celia Symons: and so overall I I've shown our work using Time series data and Edm. To quantify species, interactions and stability into these quantitative, continuous variables that could be predicted in statistical models. Celia Symons: our drivers determined by experimental and theoretical studies. So Celia Symons: these are, I think, really awesome methods to extract a ton of information from observational data that already exist.
45:01
Celia Symons: and allow us to assess how concepts where maybe we're getting different results at experiments of different scales. or in theoretical work Celia Symons: at different skills. And we can actually use these methods to test them. Enough systems, for all the messyness of nature is occurring.
45:26
Celia Symons: So the next steps for this work for me are expanding the the scope of our work. Celia Symons: So i'm part of a group that is developing a new database. The group that working on this with was formed out of the Global Lakes Ecological observatory network Celia Symons: that's led by Stephanie, a Phd. Student in Cornell. Michael Meyer is that usts now. Celia Symons: and 1 3 Dfo and Canada. Celia Symons: And this is a large group of people who have put together
45:58
Celia Symons: the largest freshwater plankton time series database with. as you can see, lakes that are mostly sampled at monthly or greater frequency. Celia Symons: A lot of the results I showed you today suggest that seasonality is really important in our lake systems. Celia Symons: and so i'm particularly excited to add more tropical lakes. Celia Symons: more southern lakes. To the analysis, I think
46:21
Celia Symons: a little bit more on a little bit more data insight into what happens when we just become warmer and system ship from something like a northern temperate lake to a northern hot, or even tropical hot lake.
46:35
Celia Symons: And we're also interested in going beyond BC's interactions and predictability Celia Symons: to focus on how species influence ecosystem services Celia Symons: or ecosystem function. Celia Symons: So, for example, I am particularly interested in using some of the same modeling techniques on species trade, the ecosystem function relationships, and see if we can determine how to create scale up to influence like functioning through time.
47:04
Celia Symons: Okay, so with that i'd like to thank my sources of funding Celia Symons: my contributors to this project in particular, Tony Robbers, who has done a lot of the analytical work here, and many sources of late data. And I thank you also, listening.
47:27
Zoe Rand: Thank you, Celia. That was great. We have some time for questions, so if anyone has any questions you can feel free to write them in the in the Q. A. If you're online and you want to raise your hand, and that I can unmute you If you want to just ask your question. Zoe Rand: Yeah. Zoe Rand: I will let me give everyone a minute to Zoe Rand: come up with their questions. But thank you so much. So that was a really great talk.
47:55
Zoe Rand: Yeah.
47:57
Zoe Rand: So it's a Trevor branch here, and i'm kind of curious with this kind of big synthesis work where you, putting together data from all sorts of different people. Zoe Rand: How did you manage that big sort of synthesis, part of the project of getting the data together. Zoe Rand: How hard was that? That's a great question. Celia Symons: I think I would classify it as surprisingly hard.
48:27
Celia Symons: The I pulled up this slide because this is actually a much larger scale project than the 12 likes that we pull together. I believe there are.
48:40
Celia Symons: I believe, there are 246 lakes. and this Celia Symons: this database that we're putting together. Celia Symons: And so there's been years now of work coming up with ways for people to share their data in the consistent format. But then we have to actually go through each data file and actually Celia Symons: check it and put it into the database one by one. Celia Symons: And so it's been a really slow process, a lot of
49:09
Celia Symons: difficulties, making data talk to each other and figuring out what it means when there's missing data. or when there's a 0 that doesn't quite make sense. All of those things have been Celia Symons: a pretty large challenge, but I would. I would definitely like to shout out, and Michael Meyer for figuring it out for this larger scale project. I thought it really nice workflow for putting together Celia Symons: this huge database, and if anyone's interested in talking to them about
49:35
Celia Symons: kind of what they've learned. I think i'm hoping that they'll write a paper on the process of the Celia Symons: as an outsider who is just contributing to some of the data cleaning. I think it's gone really smoothly.
49:46
Zoe Rand: Thank you. hey? It looks like Sam has a question. So same, i'm gonna let you talk hopefully. That works. Celia Symons: Hey, Sam?
50:17
Zoe Rand: Sam, are you there?
50:22
Zoe Rand: Oh.
50:29
Celia Symons: I know
50:32
Zoe Rand: it says you're not muted, but we can't hear you. Maybe if he just tumps it out.
50:44
Zoe Rand: Might need to type your question out. Sam. Sorry?
51:16
Zoe Rand: Well.
51:19
Zoe Rand: does anyone else have any questions
51:33
Zoe Rand: right? Well, it looks like there might not be any questions, so thank you so much, Celia, and thank you. Everyone for coming, and we'll see you next week for another quantitative summer. Zoe Rand: Oh, we got it.
51:50
Zoe Rand: Can you read it, or would you like me to read it? I think everyone will see it. Celia Symons: Oh, everyone, is it? Okay? So let me Celia Symons: read it. So Celia Symons: elements of time, like.
52:09
Zoe Rand: Okay.
52:13
Celia Symons: Yeah. Celia Symons: Yeah. So the question is about kind of the limits of this time, like embedding approach, and really systems, and especially in systems where the data points are at a a less frequent time, scale Celia Symons: and substantial. Yeah, I mean, I think, like there there's real limits to what we can do, based on that, especially if
52:38
Celia Symons: good to see you too. Celia Symons: especially if the time scale of the observations are not mapping up with the time scale of the interactions that you're interested in. That's been kind of our biggest Celia Symons: issue with getting data at a frequent amount of time scale. So
52:55
Celia Symons: you know, even a month, long time step between points here. You're capturing not only the direct impacts or direct interactions between these traffic levels. You're probably also capturing a lot of indirect interruptions. And so how those kind of add up to Celia Symons: equate to the interaction strength over the course of the month can be a little bit challenging, so that can be, You know, even more of a problem
53:21
Celia Symons: if you're only looking at any of my annual timescales. Of course your dynamics are happening over those time sales props. You're interested in the, you know, interaction between Celia Symons: 2 whale species or something, and then that that could be appropriate
53:37
Celia Symons: hopefully, that that'll be your
53:43
Zoe Rand: okay great. I'm: glad we got that question. And that was very interesting. Zoe Rand: If there are no other questions, then I think we will end the seminar there. Zoe Rand: Thanks so much, Celia. Hope the concerts nice out there. Zoe Rand: Thank you. Bye. Celia Symons: Right.
Here to help you learn
Sign in to keep a private conversation with your course tutor.
Ask me anything about the material…
Sign in to chat