 So I'd like to introduce Dr. Dean Pearson. Dr. Dean Pearson received his bachelor's of science in wildlife biology, his master's in zoology, and his PhD in organismal biology and ecology all from the University of Montana. He has been working as a research scientist with the Rocky Mountain Research Station, USDA Forest Service Wildlife and Terrestrial Ecosystems program since 2005, where he leads a community ecology and invasive species research team exploring the causes and consequences of biological invasions and the efficacy of invasive species management strategies. He is also an adjunct faculty member in the ecology and evolution department at the University of Montana. Dr. Pearson is going to be talking about applied simple models for decision support in conservation planning and introduction to the matrix. Good morning folks. I think in the realm of assessment this is going to be a little different perspective, but hopefully this will be a tool that will be helpful for folks. So what I want to do today is I want to introduce you to a modeling tool that we've developed. It's a very simple tool. You don't have to be a modeler to use it. It's very flexible. You can apply it to all kinds of situations. And my hope is that it will allow you to get a better perspective in thinking about the projects you're working on, the parts and pieces, how they work, actually project and predict outcomes of your intended actions in a way that helps you to, I think, look and refine, look back at your actions, refine your actions to better achieve your intended objectives and hopefully avoid unintended outcomes. And finally, it has a nice graphics interface, which I think can be helpful for communicating your objectives to end users so that they can better understand why you want to do what it is that you're proposing and get on board to support what you're trying to do. So before I jump in, I just want to recognize this is not a one-man show by any means. This is, I've been leading this effort, but it's a team of lots of people with lots of different skills that have been involved. I won't go through a list of names here, but just please recognize that there's lots of folks involved here. So we built this tool from a global perspective, trying to really deal with conservation and resource management issues that are happening at global scales all the way down to very local scales. And so I want to start out by just stepping back and talking about these global issues that we started from and why we developed this tool, and then drill down to the local scale and show how you can use this thing on the ground. So we, the modern epic we live in is referred to now as the Anthropocene and recognition of the fact that one organism is now so dominant on the planet that it is impacting, influencing, potentially threatening pretty much every other species on the planet. And that of course is us, right? And we do this through lots of mechanisms that you guys are all very aware of habitat transformation, exploitation of resources, removing species around biological invasions, pollutants going into the system. We're hearing a lot about those, climate change, and all of these things independently and interactively are now putting a tremendous amount of pressure on the biosphere in ways that we haven't seen before. And as a result of this, resource managers, conservation practitioners and so forth are increasingly pressured to try and keep things together, right? This is part of the job that a lot of you folks are doing, try and keep the biosphere intact to continue to provide the ecosystem, the resources that we need to survive and to thrive on the planet. And in doing so, we're using increasingly powerful tools out there. And I'm going to talk about some of the more, I think, intensive tools that are that are being used. And that is actually intentionally introducing species and intentionally removing species, extirpating species, for conservation purposes. Now, the stuff I'm talking about today applies to all sorts of resource management, but I'm going to really focus initially here on these really intensive practices. So these are things like assisted migration for climate change, biological control for invasive species, rewilding to reestablish ecosystem functions and systems, invasive species eradication, done more on islands, we more control them in mainland systems, right? And gene drives, finally, where we're again targeting invasive species or mosquitoes from malaria, things like that. So these are pretty powerful tools. And as Peter Parker said, with great power comes great responsibility, right? So I want to talk a little bit about the potential of these tools to do really good, have really effective outcomes, and for the potential of them to have side effects that we may not want to see. So a really good success story is biocontrol prickly pear in Australia. So in around, by around 1920 in Australia, prickly pear had really devastated the system. Let me find my buttons here, like you can see in the separate photo. And a single, well, several biocontrols were introduced, but one that was really effective was a moth species, cactoblastis cactorum. And it pretty much devastated this outbreak, which was all over Australia in about five years time. So these are before and after photos of the same site. You can see how powerful this tool was. So these tools can really have very powerful, effective outcomes when we use them effectively. But that's not always the case. And Macquarie Island is a case study where we get mixed outcomes. And so it's an interesting one in that regard. So Macquarie Island is a large island out in the middle of nowhere, it's south of Australia. Because it's large and so isolated, it has lots of endemic species there. It has endemic bird species and endemic plant species. And it's been recognized as a world heritage site for that reason. So a pretty unique place on the planet. Now in the early maritime days, sailors went around, they introduced all sorts of things all over the place. And here they introduced European rabbits. This was a food source on the ships and on the islands. And these guys, of course, began to eat things right away, the endemic plants, which was not great. But they also introduced domestic cats. And these were kept on the ships to control rats and mice that were inevitably on the ships, right? And they would get onto the islands as well. And these guys, they'd eat the bunnies, but they also eat the birds on the system. And they were actually accused of extirpating five species of birds from this island system. Okay. And in the case of endemic species, this means global extinction, right? So this is a big deal. This is a big impact. And it was proposed that the cats be removed from the system. And that was effectively done. And I should say that we're getting actually very good at extirpating all sorts of species, all the ways down to mice from island systems, all the way up to goats and things like that. And so this is becoming more common tool. Now, this was great because it got rid of the predation on the birds as with the intended outcome. Okay. So very effective in that regard. But it also removed predation on the bunnies and bunnies did what bunnies do. And they began to really decimate these endemic plant communities in the system. And as a result of that, not to stop there, there were invasive plants in the system that were not particularly thriving. But when this happened, these invasives really started to take off. So here we have a case where we have really good intended outcome, right? A really important one. But we also have some pretty deleterious side effects in this system. So this just illustrates how complex some of these outcomes can be when we're trying to manipulate these systems. So this is a single case study. But we wanted to understand how widely this was happening. We did a global literature review. And we found out in the very short conclusion that this was happening much more often than we would like to see happen out there. And oftentimes when it did happen, it was a sort of thing where in retrospect, you look back and you said, we could have predicted that. In this case, it's not too surprising that getting rid of cats that are eating the bunnies resulted in these sorts of outcomes, right? And so a big point is that we could, in theory, if we vetted these things more going forward, we could do a better job of achieving our intended outcomes and avoiding some of these side effects. And so toward that end, we dug in and did our modeling work. But I want to talk to you. I want to give you a little background before we jump into that. We want to do a little segue into Community Ecology 101 because there's a couple of pieces of information that are really important in getting where we got to and important for you to understand to actually apply these models. So when we go out and do a management action, it might be all kinds of things. But let's say we want to introduce or get rid of this ground squirrel in the system. Oftentimes what we find when we're looking at all this massive literature review that we did is that we get very focused on the one objective that we have in the system and we forget that our organism, our system, exists in a whole complex of other species, right? It's connected to all sorts of things in the system. So of course, other things are going to be affected. And really what we're saying when we want to try and improve these outcomes is we want to try and predict and understand these sort of rippling effects through the system, which seems like a really complex challenge, right? How do we do this? And it is complex, or at least it's a little overwhelming at first. But if you step back, there are really just five pieces of information we need to be able to actually predict these sorts of outcomes. Let me just walk through them. So the first thing we need, of course, we need to know the other species in the system or elements of the system. These could be abiotic components of the system, right? That might be interacting with and affected by our organism. The second thing is we need to know which ones are actually linked to the organism directly or indirectly. And then the third thing we need to know is the nature of that linkage. What do I mean by that? So if we look at this coyote and the ground squirrel here, well, the coyote eats the ground squirrel. So that's clearly negative for the ground squirrel. And it's a good thing. It's positive for the coyote. So that's what I mean in terms of the positive and the negative aspect of those linkages. Okay, so that's three of our elements right there. And that really, a lot of that stuff comes from basic natural history, basic understanding of the components of your system, right? So that's not so hard to get these. The next thing we need is abundance information. And this is important because, of course, if we have a really abundant species in the system and we start pushing on it, that's going to make big ripples in the system, like throwing a big rock in the pond. On the other hand, if we have species that are rare in the system, probably not that big of a deal in terms of how we push on them, but they're more likely to fall off the table, and this can be a problem as well. So abundance is important to understand, and we can get abundance information readily for a lot of organisms. The last piece of information that we need is interaction strength. And so this is a measure of how strong each one of those linkages is in the system. And this is important because really strong interactions, again, are really going to push things around. Weak interactions, we can maybe ignore and simplify our system and more easily wrap our arms around the system. Okay, so with these five pieces of information, it has been demonstrated mathematically back in 1974 by Robert May that you can predict the outcome of manipulating the abundance of any one node, any one organism in this network on the abundance of all the other organisms in the network is basic math. So you might be asking, given everything I've just said, if we've known this for 50 years now, and this is the key to really better understanding and manipulating our systems, why aren't we applying this tool? It's a good question. The answer to that is this. So traditionally in ecology, we use really precise quantitative models to predict outcomes. Now they can predict very precise outcomes, but they require very precise inputs. And it's really hard to parameterize these models for that reason, to get all this information and do this. They also have assumptions of linearity and equilibrium dynamics that are not often met. And so as a result, these models, in real world, they break and they make very bad predictions. And so in theory, this works, which is important to understand, but in practice, we haven't been able to get around this. So we came up against the same problem and struggled with it and tried to figure out how do we get forward. And what we stumbled upon was that there is a modeling approach, qualitative modeling, that's done in the social sciences and engineering fields that actually is doing the same sorts of things that we want to do. So here's a network right near. It's a social network. You have the nodes here. We have the arrows, the linkages, the pluses, the minus signs. We could add abundance and interaction strength here. This is very much the kind of network, very parallel to the kinds of networks that we're trying to work in, in national systems. Now what these guys are doing is they're doing the same thing. They want to know if I push on this node, how is that going to affect this whole network? So same questions, but they're asking for qualitative answers. They're saying, can I predict which node might go up or down and if it's going to go up or down a little or a lot? That's it. So just qualitative predictions. And it turns out that if you're just making qualitative predictions on the output, the bar for the inputs for your models to run your models goes way down as well. And in fact, you can use information, a range of information that can range all the way from very precise inputs where you have really exact data to actually guesses where you have a pretty good guess. Not maybe, you know, the better the guess, the better the model is going to work. But you can plug in a real wide range of information with these models and the models are robust and they actually predict reasonable outcomes. Now they're qualitative predictions. So if you don't need exact precise predictions, this is still enough to really get forward in terms of these understandings. And I would argue that you don't need precise predictions for a lot of the things that we're doing. This case study is interesting to me. This is Henry Kissinger's plan for Middle East peace displayed in this modeling format. I don't think it worked, given what we're seeing today. But still historically interesting in terms of the applications. And these models get used for in battle scenarios. Should we take out the bridge? Should we not take out the bridge? It's valuable to the enemy. It's valuable to us for weighing out these sorts of questions. Okay, so they've crept into the resource management conservation realm and are being applied to questions like, all right, we want to reintroduce Tasmanian devils. What's going to happen in the system? We need to get rid of invasive lionfish or wallabies that are really problematic or preserve threatened species. Or try and predict climate change outcomes. So they're getting used for these purposes, but the problem is they're predicting outcomes, but we don't know if they're correct or not. So what we wanted to do was we wanted to vet these models before going in farther. So we vetted them in case studies where we knew outcomes. I'm going to go through one of those. And that's Yellowstone Lake and Lake Trout introduction into Yellowstone Lake. So Lake Trout were introduced illegally by fishermen into Yellowstone Lake. And it really turned that system on its head. And because it's in Yellowstone National Park, it was studied intensively before this happened and intensively after this happened. So we have really good data. So the idea here is we take the data from pre-introduction, we parameterize our model, and then we ask the model what is going to happen when Lake Trout are introduced. So let me explain what happened in the system and then I'll tell you how the models did. So prior to Lake Trout native Yellowstone cuts were the dominant fish in the system. There was top of the food chain in the aquatic system. And as a result, they were the bottom of the food chain in the terrestrial system. So they were a major food source supporting bald eagle populations, osprey populations, other fish eating birds like gulls and pelicans, things like that, river otters, and even grizzly bears. So what happened after Lake Trout were introduced? So Lake Trout are much bigger fish. They pretty much decimated the cutthroat population. There's still cuts in the lake, but the biomass is now dominated by Lake Trout. And they're a very different fish. So they're a cold water fish very deep. They stay very deep in the lake. They spawn deep in the lake. They don't go in the creeks to spawn. And they don't hang out on the surface where they're accessible to bald eagles and ospreys and things like this to feed on. And so as a result, what happened in the system was we saw declines in bald eagle populations, in osprey populations, in these other fish eating birds. We saw declines in river otters. And we didn't see declines in grizzly bear abundance, but we saw a shift in their behavior. So grizzly bears, there weren't enough cuts in the spawning streams for them to bother with them anymore. Spawning happens at the time when elk are calving in the park. And so these grizzlies sort of focus on focusing on hunting elk calves. And to the point that they actually were reducing recruitment of elk calves in the system. So this is pretty crazy. This is Lake Trout introduction going to cuts, going to grizzlies all the way out in the terrestrial system. Okay. All the way out to elk calves. Okay. So complex interaction. So how did our models do in terms of predicting this? Well, when we gave the models all the information that we had, the models pretty much predicted everything that I just told you, including this very complex chain of interactions. Now we know in the real world, we rarely have systems that are studied as well as this. And so we dummied the models down. We took them way back to the point where they had no information on abundance of species, no information on interactions, strength of species. So just the three, who's there, what are the linkages and what are the nature of those linkages? And what we found was the model still did a pretty good job of predicting which nodes are going to change and are they going to go up or down. They did less well at predicting which abundance category that we're going to shift into. So we had five categories, very low, low, medium, and so forth. But even when they missed the target, they often only missed by one category. So instead of predicting that eagles were going to become very low in abundance, they predicted low in abundance. So and I would argue in those cases, that is still plenty good information for us to understand how the system is going to play out and to refine our management to try and address that. All right. So the models seem to work okay. So we decided, let's take all this modeling, this complex math, let's put it into a user interface and let's put some makeup on it, make it pretty and easy to use so that non-modelers can have access to this. And so now this is on, anybody can use this who has access to a computer and the internet. And you don't have to be a modeler to use this tool. And what I want to do now is just hopefully jump in here and show you an example of how you might use this tool. Okay. Excellent. So this will be a little challenging here. In a couple of hands. So this is, so if you just type in matrix, MPG matrix into the internet, you'll hit this landing page. You'll see this get started button. So here the picture is going to rotate through different models that we built in the system. You can hit this button. You'll get to this matrix gallery. So these are different systems in here. For instance, the lake trout system is right here. You can go in there and you can introduce the lake trout yourself and see the same outcomes that we, that I just explained to you. Let me go back to the gallery. And I put together a little scenario here for you that I'll walk through to show you how this can be helpful. Okay. So this is the scenario that I've created here is we have an invasive weed. Spotted napweed here is our nasty weed. I guess you can't see the bottom of this, but that's probably okay. So we have our nasty weed. So we have a node for the weed. We're representing the native plant communities by native grasses and native forbs. So a very simplistic model. Below this, you can't see there's another node for precip. Precip is obviously important for plants. Ask any rancher or farmer, they'll tell you that. We have honeybees in our model because we have end users that are concerned about the management that we want to do of this weed. Okay. Honeybees love napweed and they're quite happy with it. And so these guys are concerned that controlling this species is going to impact their industry. So we want to address this. So then we have two management tools that we have herbicide and biocontrol up here. So I want to note that these models can deal with organisms, human concepts, management actions, abiotic factors. You can integrate all kinds of things into these models to address all sorts of questions. So in this case, what we wanted to do here is we want to see what's going to happen in terms of applying these different tools. So I should, a little more background. So each one of these is our nodes. There are connections in the background, the same five elements that I talked about, interaction strengths and so forth, right? The numbers on the front are indicating the abundances. So here, these are percent covers for these plant groups. So these things differ depending upon what the category is. So here we go. We're going to jump in here and we're going to ask what happens if we do apply herbicide. So I'm going to move this bar for herbicide from zero up to, I don't know, somewhere about halfway up. And we'll see what happens here. And so now you see the numbers change color. They were black. If they're blue, they went up. If they're red, they went down. But you can also see the bars here. So let's look at the herbicide. I guess I'm blocking it. So the nasty weed. So there's a gray bar. That's where it used to be. And this is where it is now. So the nasty weed went way down, not surprising. We have a good herbicide. It's sensitive to that herbicide. We push it way down. So what happens with the natives? So the native grasses went up quite a bit here, which is great. A native forbs, the color here is blue, but we don't see the bars because they're so close together. So went up maybe a tiny amount, right? So what happened with our bees? Well, our bees went down because our nasty weed went way down and the bees really like it. And we got grasses to go up. Bees don't care about grasses, but the forbs didn't go up. So why did we get this outcome? Well, the herbicide is a broadly herbicide, right? We sprayed broadly across the community. And so even though it had a much stronger effect on the target weed, it still had a negative effect that kept the native forbs from really rebounding as well as the grasses. The grasses don't care about this broadly herbicide at all. So this is our outcome. So our beekeepers maybe are a little concerned about using this tool. So let's go back here and let's reset. Let's hit the reset button right here. Reset our system. And let's ask, what happens if we use the bio control in this system? We'll bring it up to about the same level here. Okay, so here we go. We have our weed went down again. Our weed didn't go down as much. The interaction strength for the bio control versus the herbicide is not as strong. We know this about these sorts of tools, at least in this case, as opposed to the Australia scenario. So we see the native grasses came up. You can't see these bars very well. The native grasses came up. And the native forbs also came up in this situation. So in this case, our tool that we're applying is very specific to the target weed. It's not having any negative spillover effects in the system. And so the forbs are able to respond positively. And we see as a result of that, the bees, they're showing blue, they went up, but such a small amount that doesn't matter. But they didn't go down, okay? So my point here, hopefully you can see, is that you can walk through and think about your system and how the interactions play out in your system. Not just the direct effects, but also complex indirect effects, some of which are going on right here, that are playing out. And you can demonstrate this for end users to help them understand what you're trying to do, to help them understand the pluses and minuses. In this case, you might use for the herbicide, the rancher who's got cattle might not care about the herbicide outcome. He might like that outcome because the cows like to eat grass. So bringing the grass way up and not the forbs is fine. For the sheep, folks, they want the forbs. So you might use the tool differently under different scenarios for different end users. So I think I'll just stop there. And yeah, I'll take any questions if folks have questions. Yeah, I can and you're digging right in. So in this case, I've actually done experimental manipulations and I know the interactions between the weeds and the herbicide effect and biocontrol effect in this situation. So I actually use from the literature studies that we've done to parameterize the interaction strengths and they're relativized. And that's the beauty of this qualitative modeling. So I'm just saying that the herbicide has a really strong interaction. The biocontrol has a pretty strong interaction. In some cases, a lot of cases, you can't get interaction strengths. And so what we've developed is an understanding here where if you have the other four of the five elements that I talked about, and you think you're pretty confident with those, you can basically, it's basic algebra to sort of get to the interaction strength. So you can just work with the model and iterate through the model. And alter interaction strengths until it fits the abundances that you have and the way the real world is, the way your model is reflecting the real world. At that point, the assumption is I must have the interaction strengths right. That's the only variable left. And so you can actually derive the interaction strengths, which is good because those are really hard to get. Yeah, that's a good question. You can just type in MPG matrix.