– Welcome everyone to Wednesday Nite @ the Lab. I’m Tom Zinnen. I work at the University of Wisconsin-Madison Biotechnology Center, and for the Division of Extension Wisconsin 4-H. And on behalf of those folks and our other co-organizers, PBS Wisconsin, the Wisconsin Alumni Association, and the UW-Madison Science Alliance, thanks again for coming to Wednesday Nite @ the Lab. We do this every Wednesday night, 50 times a year. Tonight, it’s my pleasure to introduce to you Grace Wilkinson. She’s a professor with the Department of Integrative Biology here, and with the Center for Limnology. She was born in St. Paul, Minnesota, and went to high school at White Bear Lake, Minnesota. Then she went to undergraduate at St.

Olaf College, and then went to the University of Virginia to get her PhD. However, she did most of her field work in the lakes in northern Wisconsin. Tonight, she’s gonna be speaking with us about are the algal blooms in our lakes actually getting worse? Would you please join me in welcoming Grace Wilkinson to Wednesday Nite @ the Lab?

– Thank you, Tom, for that wonderful introduction. And thank you also for the opportunity to share my research on algal blooms in lakes. Every summer, it seems like we’re hearing more and more about harmful algal blooms in our lakes and reservoirs. They’re disturbing drinking water sources, it’s causing beach closures, pets and livestock die, and wildlife are harmed. Now, as a person who really enjoys lakes, being out on them, boating, fishing, general recreation, this really concerns me. Of course, blooms get in the way of being able to enjoy the plentiful freshwater resources that we have here in the state of Wisconsin. And in some cases, can pose a really serious threat to our health. However, as a limnologist, or a person that studies inland waters like lakes and reservoirs, when I hear these stories and see them coming up again and again in the news, it makes me wonder whether there’s something larger going on.

And that is really the heart of this deceptively simple question that I’m gonna try to answer today, are algal blooms really getting worse in lakes? But before I can do that, we’re gonna start our conversation with talking about what is an algal bloom and what causes them. And then that’s gonna allow us to explore if algal blooms are, in fact, getting worse in lakes everywhere. And then finally, I’m gonna share with you some of the tools that we’ve been trying to develop in order to predict blooms before they happen in the lakes where they are occurring. So first, what is a bloom? Well, algae can form a bloom when they become essentially too prevalent. So algae are photosynthetic organisms. They can be single cells, or colonial, or filaments, like you’re seeing on the screen now. And they form the base of the food web. So they’re really important to have in our lakes and reservoirs and streams and the ocean, because they’re a food source for the rest of the organisms that live there. The problem comes when they become too prevalent, and that’s a bloom. So blooms are the rapid proliferation of algae to nuisance, and sometimes even harmful levels.

And what can make a bloom particularly harmful is when algae at really high concentrations, they are often dominated by cyanobacteria. And some cyanobacteria produce toxins that affect the liver and the neurological systems in our bodies. And one of those toxins, which is pictured here, is very common. It’s called microcystin, and it’s a hepatotoxin or a liver toxin. Now, at really high concentrations, these toxins can be lethal to animals, such as dogs that might be running around and splashing in the shallows of a lake or reservoir, livestock that are using a pond as a water source, and even, in some cases, to humans. But most frequently, we’re exposed to lower concentrations of these toxins, but there’s still sublethal effects that can happen. And those are things like rash, headache, vomiting, sores, and abdominal pain. And there’s lots of pathways that we can be exposed to these toxins. Of course, one of them is directly drinking the lake water or the source water that has these toxins in them. And that’s certainly a problem for our drinking water supplies.

But this can also be water that we use, for example, irrigating crops. We also often eat food out of these ecosystems, and fish, mussels, clams can have these toxins as a part of their tissues, but usually at pretty low concentrations. Of course, when we’re out enjoying our lakes and reservoirs, we might be swimming, water skiing, boating, fishing, there’s the opportunity during those activities to be ingesting some of the water that could contain cyanotoxins. And there’s even evidence that when waves break along the shoreline, that creates aerosols or tiny droplets that may contain these toxins, that we then breathe in if we’re simply going on a walk around the lake or live nearby. So while these tend to be pretty low doses, over time, they still could have pretty negative effects for our health, and are certainly not something that we wanna be messing around with. And so, when toxin concentrations are high, that can actually make it harmful to be around or in that water body. In addition to the health effects, algal blooms also negatively affect ecosystem services. They interrupt our water supply through that toxin production, but lots of algae also produce taste and odor compounds that just make it really unpleasant to drink that water. They can interrupt food sources through those toxins, as well as fish kills. They certainly impede our ability to enjoy those water bodies for recreation and can cause beach closures, and actually just overall reduce the perceived value of that lake for recreation.

They lead to a loss in biodiversity in many instances, because there are events like fish kills that happen with these large algal blooms, and can cause or help lead to the collapse of the food web. Even properties that are around lakes that experience frequent harmful algal blooms are found to have had lower property values overall. And as you can imagine, all of this loss of ecological or ecosystem service leads to an overall economic loss as well. In particular, that affects the communities that surround these lakes and reservoirs. So how do these harmful blooms form? Well, one of the key ingredients that algae need in order to grow is just like any other plant that we might find on land. And that’s a source of light and heat. So light, of course, is really important because these are photosynthetic organisms, and there’s more light up at the surface of the water than there is down deep, which is often why we find algae up at the surface, and that’s where blooms occur and where we can see them. And just like any other organism, algae grow faster and they thrive when it’s warm. And so, the warmer the water, the faster those algae will grow. Now, another key ingredient for algal growth are nutrients, and in particular, nitrogen and especially phosphorus.

Phosphorus is really important for algal growth. And those nutrients often come from the surrounding land or the watershed, and are brought into the lake. And what we do in the watershed as humans, how we maybe use that land for things like agriculture or urbanization, is gonna influence how much nutrient is coming off the land and into the lake and in what form. Now, those nutrients come into the lake, but they don’t necessarily get taken up immediately by algae. A lot of them actually end up in the sediments or at the bottom of the lake, where they can be stored for years, or even decades. So this is a legacy of the past actions that happened on land and the loading of that nutrients from the land into the lake. But those nutrients don’t stay there. Instead, they often can be released, either through chemical processes or physical processes back up into the water for algae to take them up. So for example, if there are a lot of carp in the lake, as is pictured here, carp really like rooting around in the sediments to find their food, and that’ll stir up the sediments, releasing a lot of that phosphorus. In addition, if there’s a lot of waves because it’s a large lake or there’s a lot of boat activity, that can also stir up the sediments and release phosphorous.

And how long those nutrients stick around or are available for algae to take them up and use them for growth is also gonna influence when and where a bloom can form. And so, we refer to this as residence time, or how long a nutrient molecule stays in the water column and may be available for algae to use for growth. But it’s not just algae that are growing. There is also competition for these resources by other primary producers or plants in the environment. So aquatic plants or macrophytes also need light in order to grow. And so, they’re competing with algae for that. And they’re also competing for those nutrients. In addition to that, algae are being grazed upon or they’re being eaten. So just like on land where we have, for example, grass growing and there’s organisms or animals that can come along and graze that grass, there are also grazers in lakes. And one of them is pictured here.

It’s zooplankton called Daphnia. You can kind of think about Daphnia as being like the cows of the lake. They’re really good at grazing on algae. And so they can help stop a bloom from forming. But if there’s things like cyanobacteria that are mostly around, that cyanobacteria is not particularly nutritious, plus there’s those toxins that it might be producing. And so, Daphnia aren’t always the best at controlling algae. And so, it’s this complex interaction of all these external, or outside the lake, and internal environmental drivers that can create the conditions that an algal bloom can form. And of course, human activities are causing large-scale changes to all these drivers. So for example, we’re changing and increasing the inputs of nutrients onto the landscape. This is an example from Sabo and colleagues that was recently published in 2021 that looked at the phosphorus inputs that we as humans are putting onto the landscape here in the United States.

And what this map shows are those phosphorus inputs, and the deeper the brown color, the more phosphorus that’s being put onto the land. And this doesn’t just come from agricultural practices, like adding fertilizer to the land, but it is also livestock waste, human waste, and a small portion is even coming from the dust particles that are in our atmosphere that also have phosphorus on them. And as you can see from this map, we’re not putting phosphorus on the landscape in the same way, in all places. In some regions, there’s very little phosphorus input, but for example, here in the upper Midwest and in the Corn Belt region, there’s a large amount of phosphorus being added to the landscape. And this phosphorus has the potential, then, to run off into our surface waters, our lakes, rivers, streams, and reservoirs. And so, what this map is showing is that potential for runoff of phosphorus into surface waters, and the deeper the red color, the higher the potential, or the greater amount of phosphorus that’s likely to be running off. And again, you can see that it’s pretty variable in where that phosphorus is running off, but there’s areas of higher concentration, like here in the upper Midwest. And it’s not just that we know we’re putting phosphorus onto the landscape. We do have pretty compelling evidence that it is leading to nutrient pollution in our surface waters. And so, this loading of phosphorus from outside the lake into our surface waters is referred to as eutrophication.

And the U. S. Environmental Protection Agency samples lakes and streams across the United States every five years in a large synoptic survey. And as you can see from this data here, where we’re looking at the percent of those ecosystem samples, this is usually well over 1,000 streams and lakes across the United States, that the percentage with a total phosphorus concentration above this 10 micrograms per liter threshold has been increasing subsequently over the past decade to decade and a half, to the point where we expect where almost all ecosystems that are sampled are likely to be above this phosphorous threshold. So this is strong evidence for widespread eutrophication of our inland waters. At the same time, we know that the climate is changing, and it’s rapidly warming, and that’s also causing rapid warming in the surface of lakes around the globe. So as you can see from this map here, every dot is a lake that’s had long-term monitoring on it for surface water temperatures. And the deeper the red color, the stronger the warming trend has been. So the majority of lakes are warming, and warming pretty strongly. And in addition to creating warmer conditions, which of course, algae love to grow because it’s nice and hot, it’s also changing the physical environment, or the way that that water is mixing from the top to the bottom of the lake, and whether or not it’s stratified.

And that physical climate is also really important to algae. In addition to, of course, the climate warming, precipitation is changing, and in particular it’s intensifying, or the big storms are getting bigger. So this map here from the National Climate Assessment shows areas where precipitation has either been increasing in green or decreasing in the brown color. And in some areas like in the desert Southwest, you can see that intense precipitation has been declining, but again, here in the upper Midwest, in the Corn Belt region, we see that precipitation is intensifying. When we look at the entire United States from the 1900s to the 2000s, it’s very clear that these big storms are becoming more common, and the big storms are getting bigger. And that is definitely something that we are seeing here in Wisconsin, right? So just zooming in a little bit more from the U. S. down to Wisconsin here. These are data from the Center for the Climate Research at the Nelson Institute here at UW-Madison. And what I’m showing with the change in annual precipitation here is the darker the color, is the greater increase in precipitation that’s been observed over the past 70 years.

And every place where there’s an asterisk, that’s a region where there’s been a significant increase in precipitation. So as you can see, particularly in the southern portion of the state here, we’re seeing a significant increase in precipitation on an annual basis. And that’s paired with an increase across the state in minimum temperatures. So minimum temperatures are important in thinking about the way that we experience winter, right? So winter, of course, is an iconic part of the Wisconsin life, but winter is changing. And with minimum temperatures increasing, we can expect to see a loss of ice cover in some of these lakes, as well as how long that ice is staying on, which changes the way the nutrients are flowing through these systems and the light that’s available for algae to grow. So with all of these things combined, we’re seeing an increased input of nutrients onto the landscape, more nutrients being stored in the sediments of lakes and being available to be released, changing warming atmosphere and changing stratification, and coupled with intensifying precipitation, it is a very reasonable hypothesis that blooms are getting worse in lakes everywhere. And coincident with all of these changes is our growing awareness of algal blooms in the news. So for example, this graphic here shows local news reports of algal blooms over the past 10 years in the United States. Every time there’s a local news report, it’s recorded and added to the graphic. And then that builds up over time.

So you can see all the places where blooms are being talked about, and talked about in the local news. So it would seem reasonable, coincident with all of these changes, and what seems like either our growing awareness or the fact that blooms really are getting worse, that we might think algal blooms are getting worse everywhere, but we haven’t yet had the opportunity to really challenge that idea or test that hypothesis with data. And that’s what my collaborators and I set out to do. Before we could answer this question, though, we first needed to define what does it mean for a bloom to get worse, right? So can we actually create some metrics to quantify what we mean by worse? And the first metric we came up with was a change in bloom magnitude. So this is the average bloom getting worse from year to year. So this might be tied to, for example, an increased likelihood that the ecosystem services that I discussed before might be lost from the system. Another metric that we identified was severity. So is the peak of the bloom getting worse and worse? In other words, when a bloom happened in the lake in the past, did it go from being crystal clear blue in the spring to kind of green later in the summer? And now, it goes from crystal clear blue to green and scummy, chewy water, right? This big algal bloom. So a more severe bloom. And of course, the greater the severity, the increased likelihood of severe consequences with that bloom.

So things like fish kills, or really high toxin concentrations being produced. And the final metric is duration, or is the bloom lasting longer and longer each year? In other words, is the bloom, instead of just being a quick, severe peak of a lot of green algae, is it sticking around for more of the summer, even the whole summer? And if so, that’s likely an increased period of summertime economic loss, particularly for the tourism community surrounding these lakes and reservoirs. So with these three metrics, we now have a way of looking at the average bloom, or its magnitude, the severity, or the peak of the bloom, and how long the bloom is lasting in order to start answering the question, are blooms getting worse? But now that we have our metrics, we need an actual variable or data in order to address this. And so, there are lots of different ways that you could go about trying to quantify how bad a bloom is in a lake, or try to quantify how much algae is there. You could do it visually. You might be able to take a picture and assess something about the color of that photo, but a lot of these metrics are subject to human biases. And so, we wanted something that would be less biased and more reproducible. And so, what we turned to was measuring chlorophyll a concentrations in lakes. Now, chlorophyll a is a pigment that’s found in algae, it’s found in tree leaves and grasses. It’s the thing that makes them green.

And so, we can take a water sample and extract the chlorophyll a in the algae that’s in that water sample out of it in the lab, and then measure it with a spectrophotometer or a fluorometer. And that creates this reproducible, quick, and cheap way to measure algal biomass. And it is, in fact, so easy to do this, that measuring chlorophyll a is a very common variable that’s a part of water quality monitoring programs across the United States. And so, this would give us the opportunity to look at lakes in a broad area to try to address this question. So now that we know what we wanna measure, magnitude, severity, and duration, and how we’re gonna measure it, chlorophyll a, we needed to know how many measurements happened each summer in order to really capture a bloom. And to illustrate what I mean here, I wanna show you some data from Swan Lake, which is located in western Iowa. And these were data that were collected by David Ortiz, a master’s student in my lab at the time. And what David did is he had a sensor that was in the water measuring chlorophyll concentrations. So unlike taking that water sample back to the lab and extracting, this was a special sensor that sat in the lake and made those measurements in real time. And so, he essentially had a measurement of chlorophyll concentrations every single day in the lake during the summer of 2018.

And you can see those data here. And so, if we wanted to calculate the magnitude of the blooms that occurred, the severity, and the duration, we could do that with these data. And since he has a measurement for every single day, we have a pretty good idea that we’re capturing sort of the true measure of the magnitude, severity, and duration. But at the same time, the Iowa Department of Natural Resources has a long-term water quality monitoring program, where they visit all the lakes in the state three times during the summer to also measure chlorophyll concentrations. And so we wanted to know, well, how do the three measurements that the Department of Natural Resources made match up with the everyday data that David was collecting? And so, if we compare those three sampling dates, which I’m showing with the green dots here, to the full summer value, you can see that these metrics that we’ve calculated, you start to get a very different picture of what this bloom looked like, whether you visited three times or every day. So we did this exercise, essentially, for 30 lakes, where we had daily chlorophyll measurements during a bloom. And we wanted to see how frequently measurements needed to be made in order to capture these bloom metrics. And what we landed on was essentially, every 14 days. We needed a measure of the chlorophyll a concentration about every 14 days to adequately capture the magnitude, severity, and duration of a bloom. And it turned out, this was probably the key sticking point in finding data to answer our question, because it’s very common for water quality monitoring programs in states to go out and visit a lake once, twice, or three times during a summer, not necessarily every 14 days.

So in order to answer our question, we needed at least 10 years of data within the lake, because we wanted to look at long-term trends in algo biomass. We couldn’t have really any large data gaps during those 10 or more years. And we needed the measurements every 14 days. And so, in my mind, these sort of became unicorn data sets. They were mythical, or something that were gonna be really hard to find, but ultimately, we ended up finding 323 lakes that met these criteria, and they span the 17-state region that I’m showing you here. Every dot is one of the lakes. Now, the majority of these data came from the LAGOS database, which was published by a group of limnologists led by folks at Michigan State. This is a fantastic resource. And really, what it did is it brought together all the data sets that different state agencies, local watershed associations, and university researchers have been collecting for decades, so that we could start asking these bigger-picture questions. And so, we’re certainly indebted to the folks that did all the work of collecting these data.

And then the folks at LAGOS that brought these data sets together. So with these 323 lakes that have the right data requirements, we can now ask our question, are algal blooms in lakes really getting worse? As a reminder, we wanted to look at the magnitude, severity, and duration, and how those were changing over time in each one of these lakes. And so, what I’m showing here is a histogram for each one of those metrics, and it has the standardized trend coefficient. So in other words, if the trend was positive for magnitude, that would mean that the magnitude was getting worse, or the average chlorophyll concentration’s growing higher over time. Or if the trend is negative, then that means that the average concentrations for magnitude were actually getting better. And so, what you can see here is that for magnitude, severity, and duration, there’s actually a really large spread in how the lakes were responding or what their trends were over time. We have lakes that were both increasing and decreasing in magnitude, having smaller or larger peaks over time in severity, growing shorter or growing longer in terms of duration. Now, our original hypothesis was that blooms were getting worse everywhere. And so, we would expect to see these distributions not look like this, but instead, be all on the far side, or in other words, on the side of magnitude getting worse, severity being larger, and longer duration. And that’s certainly not what we’re seeing.

And so, with these data, we aren’t finding any evidence that algal blooms are intensifying everywhere. In other words, there’s no widespread intensification of algal blooms based upon the 323 lakes that we looked at. But what really surprised us when we dug down and tried to understand which lakes were changing significantly over time in either direction was that 11% of lakes are getting significantly worse. In other words, one in ten are seeing their algal blooms get worse, but shockingly, 17%, or approximately one in six, are getting significantly better. So more lakes are actually, in our data set, getting better in terms of algal blooms instead of getting worse. And this was certainly a surprising finding for us, and very much did not fit with our original hypothesis. And so, of course, when you get a surprising result like this, the first thing you wanna ask yourself is, why? Why are so many lakes getting better? And I think it would be important to note that, while we were approaching this question using chlorophyll measurements that had been made in lakes over decades, often collected by volunteers and such, there are others who have been approaching this question from a different perspective. And in particular, Topp and colleagues published a paper in 2021 where they used remote sensing or satellite measurements to look at water clarity. Now, water clarity is certainly related to algal blooms. When there’s a bloom, the water’s not particularly clear, but there’s other things that affect water clarity as well.

So it’s a little bit different of a measurement. That being said, they looked at 14,000 lakes across the United States and found that, very similar to our results, more lakes were becoming clearer than were becoming less clear. So while our evidence and our analysis was certainly surprising to us, others approaching the question as well are finding similar results. And this builds a weight of evidence, or threads, that perhaps this might be the actual picture of what’s happening out in the United States. And so, we wanted to know, why? What might be happening at sort of this broad spatial scale, and why are so many lakes getting better? And one hypothesis that we have is that the Clean Water Act and other water quality policy, we might actually be seeing the benefits of that policy, which was enacted decades ago, actually starting to play out. Now, this is really hard to test. There’s no federal database of lake restorations. There’s not even state databases of lake restorations. So finding information on whether or not a lake has had restoration activity is surprisingly difficult. What we were able to do though is look at local news reports, which reported on whether or not a lake had received restoration activity.

And that’s because if the state is about ready to spend a couple of million dollars, or a local lake association, they wanna put the word out and put out a press release, and it’s covered in the local news. So based on that local news reporting and what we could track down, 65% of the lakes that had a significant improvement in one of the algal bloom metrics had some sort of restoration activity that had occurred during the period of measurement. So it’s certainly possible that this is a effect of policy and our restoration efforts in order to turn around bad water quality and try to address it in individual lakes. But it’s still too hard at this point in time in order to really track down if that’s what’s happening. And we can’t put these data into a statistical model in order to test that. So right now, it just remains a hypothesis. And you might be thinking, “Well, lakes that are nearby each other might be acting or reacting more similarly,” right? Their bloom trends might be similar. And it’s more a difference between regions or states. And we certainly thought that as well, which is why we took a look at this. And just to illustrate, I wanna zoom in again on the state of Wisconsin, just to illustrate how lake-to-lake differences in bloom trends were quite severe, as were regional differences.

So for example, down in the southern portion of the state, in Madison here, we have Lake Mendota and Lake Monona. Now, for decades, researchers at the University of Wisconsin-Madison have been studying these two lakes. And most recently, the North Temperate Lakes Long-Term Ecological Research program has been studying these lakes. So we have wonderful chlorophyll time series or data sets to answer this question. And in Lake Mendota, we’ve seen a significant increase in bloom magnitude and bloom duration during the past couple of decades. But surprisingly, right next door, just across the isthmus, Lake Monona, we’ve seen no change in bloom magnitude, severity, or duration. Now, if you’ve ever visited Monona in the summer, you know that there are definitely algal blooms. So remember, this is trends. This is not if the lake is blooming or not, but if there’s been a change in the magnitude, severity, or duration of those blooms. So lakes right next door to each other, like Mendota and Monona, but in other places in our data set as well, are responding very differently.

And there’s certainly differences across regions as well. So if we compare the southern portion of the state to the northern portion, like Allequash Lake up in Vilas County, we see that blooms have been actually getting better in this lake. In other words, the severity and duration of blooms has been significantly declining over the same period that Mendota and Monona were studied. So we wanted to dig in more into these lake-to-lake and regional differences and try to understand what’s driving the difference in bloom trends among lakes. And we have five hypotheses. Could be related to the lake morphometry, or the shape of the lake, because that influences the way that the sediments and the water interact, and whether that water column is mixed or stratified. Could be the land-to-lake connectivity. So where that lake is sitting in the landscape, whether it’s fed by groundwater or mainly surface water, and what the watershed looks like, what are the land uses and potential nutrient inputs? We also looked at nutrient availability. So just how much nitrogen and phosphorus are in the water. And has that been changing over time? We also wanted to look at whether climate mattered.

In particular, whether warming temperatures during the summer or any trend in temperature during the summer was related to algal bloom trends. And we looked at whether or not big storms, so those really intense precipitation events have been increasing or decreasing in the watershed of these lakes, and if that helped us understand these bloom trends. And in short, we really didn’t find any evidence that temperatures were related to trends in algal blooms, which was somewhat surprising, because we expected, given this large-scale increase in global temperatures, that that might be playing somewhat of a role. Similarly, we didn’t find any land-to-lake connectivity metrics that were related to the algal blooms. Instead, we found that large lakes were more likely to have increasing trends in bloom magnitude and severity. So larger lakes are slightly more likely to be getting worse in terms of algal blooms. Lakes with really high average chlorophyll concentration. So these are the lakes that are already pretty green, are experiencing blooms on a regular basis, were more likely to actually have a negative trend in magnitude and severity. And this might seem a bit counterintuitive, but it might actually be more evidence towards the idea that the lakes that already have pretty bad algal blooms, we’re putting more effort towards restoring. And so, those are the lakes that are getting better.

Again, it’s only a possibility, and we need more data to really test that in the model. And then finally, we see lakes that are experiencing more intense storms during the summer are also seeing a negative trend in bloom magnitude. While these individual pieces or hypotheses are certainly interesting and provide some good insight into what lakes are changing, and perhaps a little bit about why, it’s really the interaction between some of these variables that was the most interesting. And in particular, the interaction between how green a lake already is and how summer storms are changing. And to illustrate this, I’m gonna turn back to Lake Mendota. So Lake Mendota, as we know, is a very nutrient-rich ecosystem. It’s already experiencing blooms, and it’s very green in that regard. Now, if Lake Mendota were also experiencing a decline in storm intensity. In other words, those summer storms were becoming less intense or less frequent, then the statistical model tells us that we would expect a decline in bloom magnitude. On the other hand, if Mendota is a very nutrient-rich ecosystem, we’re experiencing an increase in storm intensity, we would expect an increase in bloom magnitude based upon the population of lakes that we studied.

And of course, this is, in fact, what Mendota is experiencing. In the southern portion of the state here, we’re seeing significant increases in annual precipitation, more intense summer storms, and Mendota is definitely increasing in bloom magnitude in a significant way. And so, how do we use these pieces of information to think about managing algal blooms in the face of human and climate-driven change? Well, it’s really hard because of these complex interactions between what’s already happening in a lake and what’s happening with the storms and their intensity of those storms around the lake to create sort of one prescriptive solution for all lakes in North America or the world. But we can use this information to identify those lakes that may be more susceptible to increasing algal blooms or worsening algal blooms. And in particular, in those nutrient-rich water bodies that are experiencing more intense precipitation, thinking about the importance of managing for erosion and nutrient runoff during high-energy flows. In other words, these really large storms bring a lot of water all at once. And that water has a lot of energy behind it, as it runs off the land and into the lake. And that has the potential for high erosion, as well as nutrient runoff. And so, the possibility to manage for those high-energy flows might be really beneficial in those lakes that are experiencing more intense precipitation. Similarly, for those lakes, these large storm events often bring a lot of wind, and that can cause the sediments to be stirred up, releasing that internally stored or legacy phosphorus.

Also, if there’s a lot of boat activity, or just general, it’s a large lake, and so, there’s a lot of waves, that’s increasing sediment resuspension and releasing those nutrients. And so, any management activities that help to decrease that sediment resuspension may be beneficial. But I think it’s important to note that managing nutrient-polluted lakes under these increasingly intense precipitation events that we’re seeing, if we use the tools that we currently have in our toolbox and what we’re doing now, it might only be managing to run to stay in place. In other words, we’re gonna need to put in more effort and more management and restoration in order to overcome increasing precipitation intensity. So while we’ve been able to demonstrate that algal blooms aren’t getting worse in the majority of lakes, and in fact, in many more, are getting better, there certainly are lakes that are still experiencing algal blooms every summer. And so, this other portion of my research that I want to share with you is related to tools we’ve been developing to predict algal blooms before they happen. So what I mean by that is, what if we could predict a bloom during this pre-bloom or blue phase that I’m showing here, and before we could actually see it, when the bloom’s occurring? So imagine we have one of those chlorophyll sensors, and it’s sitting out in a lake collecting data every day on the chlorophyll concentrations. And during this pre-bloom phase, that’s when we’re out using the lake, because it’s gorgeous, it’s nice weather, we’re swimming, boating, recreating, perhaps it’s a drinking water source, perhaps it’s a source where people fish for food. And during this period of time, there is some changes in the chlorophyll concentration, but they’re not large, and it’s certainly not bloom conditions. And what we wanted to know is, can we use this information during this pre-bloom phase in order to predict when a bloom is going to occur with the goal of perhaps protecting or mitigating, or even trying to prevent some of these large consequences that a bloom can have on both our natural resources as well as our health? And so, one way that we can do that is using theory from ecosystem science about critical thresholds and regime shifts in lakes.

So this is the same diagram that I was showing before. And I want you to consider the macrophytes or the aquatic plants that are growing there. Aquatic plants are, as I mentioned, competing with those free-floating algae that form the blooms constantly for light and nutrients. But at some point, with nutrient loading from the land, or being released from the sediments, there’s a critical threshold at which those algae overtake aquatic plants and form a bloom, and they end up growing. They win the competition. And it’s this regime shift from a clear water aquatic-plant-dominated to an algal-dominated system that we’re trying to predict. So what is a regime shift? Well, it’s a sudden transition from one stable state in a system to another. And that transition is caused by a loss of resilience. Now, this isn’t just something that happens in lakes. It happens in many ecosystems, like grasslands converting to shrubland, salt marsh to tidal marsh, corals to algae-dominated dead coral systems.

These even happen in the human body preceding seizures, and in economic markets. Now, the thing about these is that they’re sudden transitions. And so, they tend to come by surprise, and that makes them seemingly difficult to predict. But given the consequences, which can be pretty severe of these transitions occurring, it’s necessary to be able to predict them to protect ecosystem and human health. And so, it’s this transition, again, in particular, that we’re focused on, from the clear water state, where aquatic plants are dominating, to this algal-dominated turbid state that you’re seeing in the photo here. So to better understand how we can use this regime shift to predict algal blooms, imagine that these two ecosystem states are represented by basins, or the cups that you’re seeing here. So the blue clear water state and the green algal-dominated state. And the lake is the ball that’s currently sitting in the clear water state basin. Now, there’s a really large saddle or hump between the clear water and algal-dominated state. And that’s because there’s really high resilience in the system.

And that resilience is largely dictated by the amount of nutrient loading that’s currently occurring. So in this case, resilience is high, the lake is in the clear water state, and nutrient loading is quite low. But we know, as we just talked about, nutrient loading is not something that remains constant. It’s pretty stochastic. And so, something like a storm can happen, and that’ll be like a perturbation that is adding a bit more nutrients, all of a sudden, into the system. And you can think about that disturbance or that perturbation as pushing the ball around in the basin. So the ball’s moving around, but because the basin walls are so steep, the ball doesn’t move very far, right? There’s not a lot of energy behind that disturbance. And so, if we were to be measuring, for example, chlorophyll a concentrations in the lake over time, we would see that when these storm events happen, there’s some variability because the ball’s moving around in the lake, but that variability is quite low, because the resilience is high, the basin walls are really steep. Similarly, another metric, autocorrelation is low, and if you’re not familiar with autocorrelation, you can think about it as, how much does today look like yesterday? And so, again, because the basin walls are steep, things are just sort of variable, but in a small way. So autocorrelation is quite low.

Now, if nutrient loading over time begins increasing in this system, perhaps there’s a change in the way land is used in the watershed. That same small perturbation, that storm event is gonna come along, add some more nutrients, and the ball is gonna be rolling around in that clear water state basin. But because nutrient loading is higher, resilience is lower, the basin walls aren’t as steep at this point. And so, that same small push is gonna push the ball farther and farther away from the deepest point of the basin. And so, if we’re still measuring chlorophyll concentrations in the lake during this time, we would see that the variance or the day-to-day variability has really started to increase, because the ball’s moving farther and farther away, and autocorrelation is increasing. And so, these are two things that we would expect to have happen, an increase in variance and an increase in autocorrelation when the resilience is being lowered in the system. Now, eventually, nutrient loading is gonna be so high, that same small perturbation or storm event is gonna come along, and it’s gonna push that ball from the clear water state into the algal-dominated state, and we’ll be having an algal bloom. And so, what my collaborators and I wanted to know is, given that we would expect this rise in variance and autocorrelation before an algal bloom appeared, can we use these statistical indicators as early warnings of the bloom? In other words, can we measure chlorophyll concentrations regularly in a lake that we think is about to have a bloom? And do we see that variance and autocorrelation rise in a reliable way, so that we could use it as an early warning signal that the regime shift or the bloom is about to happen? And so, we tested this idea by doing multiple whole-lake experiments up in northern Wisconsin. So the lakes that we are experimenting on are pictured here. They’re Peter, Paul, and Tuesday Lakes, and they’re located at the University of Notre Dame Environmental Research Center up in northern Wisconsin and the Upper Peninsula of Michigan.

Now, from 2013 to 2015, we slowly added nutrients to Peter and Tuesday Lakes with the goal of trying to induce a regime shift, or make a bloom happen. And at the same time, we were monitoring Paul Lake, and that was our reference system. So we weren’t adding nutrients. We just wanted to make sure that the variability that we were measuring, that rise in variance and autocorrelation, was really related to a bloom and not something else happening to these lakes. So that’s why Paul was the reference. So every day, we would go out and add nutrients to these lakes, and we’d be measuring chlorophyll concentrations in the water using sensors like the one that’s pictured here in the center. And even if we saw these statistical warning indicators arise in variance and arise in autocorrelation, we’d continue to add the nutrients to make sure that we were actually seeing an early warning of a bloom occurring. So just as an example, here is an aerial view of Peter and Paul Lakes. Peter is the green lake. Paul is blue lake that has a bit of clouds reflected in it.

So that’s the white that you’re seeing there. And no surprise, when you add nutrients to a lake, you can induce an algal bloom. And so, this is a bloom from 2014 in Peter Lake that I’m also showing you the data for. So this is a time series with Paul shown in blue of chlorophyll concentrations that were taken from that sensor. And you can see that really, there was no change in Paul. We weren’t adding nutrients. There wasn’t a bloom. But in Peter, we added nutrients and slowly over time, we built to a really large algal bloom. But our question was, were there early warning signals prior to this bloom occurring, a rise in variance and a rise in autocorrelation? And in order to answer that question, I just wanna briefly describe how we measure a rise in variance and autocorrelation. So we have the sensor sitting in the water, and let’s say it’s made measurements for 14 days.

So we can take those 14 days’ worth of data, what’s shown in that green box there, and calculate some measure of variance, say the standard deviation. Then another day arrives, more measurements are made, and we roll that 14-day window forward, and recalculate standard deviation, and continue doing that with each day’s data coming in. So in that way, we can look to see if variance or autocorrelation are increasing. But it’s really hard to say what actually constitutes a rise in variance or autocorrelation, and what’s just sort of natural variability over time. And so, we used a mathematical method called the quickest detection method that lets us know when a threshold has been passed in variance or autocorrelation, and essentially it sounds an alarm. So every day, we calculate the quickest detection metric and ask, is a bloom coming? Is there an early warning alarm? And the metric essentially tells us yes or no. So we did this, as I mentioned, in Peter and Tuesday Lake from 2013 to 2015. And so, how many times did we see these early warning alarms happen before a bloom? Well, we saw over these six lake years of experiment between 6 and 57 days of warning before the bloom occurred. So in some years and in some lakes, only six days, or about a week’s worth of early warning notice. But in other instances, we saw almost two months of early warning.

So we knew that that bloom was coming. That’s a pretty good early warning system. Surprisingly, also, we found that the more severe the bloom, the more types of alarms were firing off. In other words, we had more confidence that really that bloom was coming, because lots of different ways of measuring these early warning alarms were firing off and saying, “Yep, here it comes. ” So the more severe, the more alarms, and the more confidence we could have. So we were really excited that this theory about regime shifts actually played out in our experiments, and maybe could be used as a tool for an early-warning system. And so, that was the next question that we asked, do the early warning signals come in time to prevent the transition to a harmful algal bloom? In other words, if we wanna maintain really high or good water quality, we know from our previous experiments that, as we get closer and closer to the regime shift, our statistical warning indicators are gonna start firing off. If we see those early warning indicators alarming or firing off, can we do some sort of intervention to pull the system back from the brink and maintain good water quality? And so it was this idea that we tested in 2015 in Peter Lake. This time, same experimental setup, but when we saw those early warning signals firing off, we stopped adding nutrients to Peter Lake. And we wanted to see if we could prevent a long-term or sustained bloom from happening, or prevent that full regime shift.

And so, at the beginning of the summer in 2015, as you can see from this aerial image here, Peter and Paul Lake looked very similar to each other. And in the graphs that I’m showing above that, chlorophyll, as well as phycocyanin, the values were very similar. Now, chlorophyll is that variable we’ve been talking about, and phycocyanin is another pigment that is mainly just found in cyanobacteria. So it’s a good way for us to measure essentially, that potentially really harmful component of the bloom, the cyanobacteria. So they were similar to each other when we began adding nutrients. And when we saw those blooms fire off, we turned off the nutrient additions and we stopped. And so as you can see, there was a bloom that occurred, a very short one that was small, in Peter Lake. And very quickly, it began to dissipate and die back. And the lake was back to being in its clear water phase by the end of the summer, and looking just like Paul. So while we didn’t prevent any bloom from happening, we certainly stopped a permanent regime shift from happening, and were able to pull that system back from the brink a bit.

So what did we learn? Well, we learned that the intervention, or turning off the pipe of nutrients, actually could have happened faster, because we had an issue with the sensors and the way that they were relaying data. And while that was certainly disappointing for our experiment, I think it’s actually pretty typical of what happens in the real world when you’re trying to collect data and make decisions in real time. But what we did find was that essentially turning off the pipe of nutrients worked, but this is experimenting in an ideal world, right? I don’t know of any lake or reservoir where there’s one pipe of nutrients going into it that you can change or turn off in one day and make a management decision. And that’s really fair. Part of the idea here was that, in this idealized setting, if we couldn’t make it work, it wasn’t gonna work in the real world. But given that there was some success, then we can think about where and when might this be a tool for at least some sort of early warning, or at least to warn people that something bad, a bloom was about to happen, and to protect their health and our natural resources? And so, the next step was finding out whether this really worked, this early warning system, not in an experimental setting. And so, that was work that occurred with the Iowa Department of Natural Resources on these four lakes here, and was also work done by David Ortiz in my lab. And as you can see, for these four lakes, they were monitored using those chlorophyll sensors, much in the same way that we were doing for our experiments. And they experienced blooms every year. These are very nutrient-rich lakes.

And what David found was that we saw pretty similar results to what we had in our experiments. The early warning indicators started firing off on average 17 days before bloom conditions were reached. And 51 days, on average, before the peak of the bloom or the most severe part was reached. So surprisingly, not even just in the experimental conditions, but also under more real-world conditions and all the variability and issues with data collection that can occur, we were still able to identify these early warning signals of harmful algal blooms. So how might this be helpful or effective? Well, I would argue that maybe the real purpose of a tool like this would be to use it as a potential for proactive management to protect ecosystem services. So those services, like I mentioned, like drinking water and food supply being protected, our irrigation waters being protected, and protecting recreational users of these water bodies that might experience a bloom. What the early warning system could provide in these instances would be opportunities for managers to know when they should dedicate their resources, time, money, and personnel towards more intensive sampling if a bloom is coming on, perhaps closing beaches, or even doing other management interventions, like turning on aerators or algicides in really extreme cases in order to prevent things like fish kills, as you see here. And in particular, I think it’s this knowing where and when to put resources, because they’re limited. We don’t have all the money or time in the world to measure everything everywhere. And if we know that a bloom is about to come on, that’s a good indication that more intensive monitoring of, say, toxins should be occurring in order to protect drinking water sources and recreational users.

So while we found that algal bloom intensification is currently not widespread, we know conditions are gonna continue to change. Eutrophication will continue to accelerate in some regions. Precipitation will also continue to change. And while we have tools like the early warning system that we can use in a management context to protect health and human resources, they’re not a solution. They’re a stopgap measure while we work on the solution. In order to protect our lakes, we need to continue to reduce eutrophication, promote active and science-based restoration, and work to adaptively manage our inland waters in the face of a changing world. I wanna acknowledge my collaborators on both of these projects, as well as our funding source, which was largely the National Science Foundation. If you’re interested in more details about these projects, you can reference the papers that are here. And I wanna thank you for letting me share my research with you today. And I hope it gave you an opportunity to view the lakes that you love in a slightly different way, and think about the actions that we all can take to protect our freshwater resources.