Speaker Info

Tom Ihde
Fisheries/Ecosystem Modeler
National Oceanic and Atmospheric Administration

Email: tom.ihde@noaa.gov


Dr. Ihde is a marine and estuarine fisheries scientist with 20 years of experience working in the Chesapeake Bay system. His main interest is in the application of the best tools and approaches to improve the management of our marine fisheries resources. In the Chesapeake, we have historically been challenged by over-enrichment of the system, habitat loss, and overfishing. More recently, climate change has increasingly been recognized as an additional, critical stressor on the system. Over the last seven years while with the Chesapeake Bay Program (CBP), with the help of a wide range of regional scientists, Dr. Ihde has built and applied an interdisciplinary, whole-system production model known as Atlantis. He uses the approach to forecast the productivity of our living resources in the context of both our restoration efforts, and the multiple, simultaneous stressors affecting the Chesapeake Bay. He is an active member of the CBP's STAC, as well as both the Habitat-Goal Implementation Team and the Fisheries-GIT.

More info...

Seminar Abstract

There have been numerous indicators that there has been progress in improving the water quality indicators for the Chesapeake Bay system. Less clear, are the effects of these water quality improvements on the living resources of this system, or the scale of those effects in comparison to other stressors like changing water temperatures in this region. Similarly, the cumulative effects of multiple, simultaneous stressors in combination with the water quality improvements are difficult to estimate, and there are relatively few tools available with which to accomplish this task. One such tool is an"end-to- end" or "full-system" modeling approach called "Atlantis". The Chesapeake Atlantis Model (CAM) is a deterministic, biogeophysical, production simulation model of the Chesapeake system, designed to provide strategic information on the trade-offs of different management choices, e.g., targeted restoration, ongoing habitat loss due to sea level rise and shoreline hardening, water quality improvement, etc. The spatial modelling approach includes a wide range of system features, the most important of which are: physical forcings of heat, salt, and water movement; predator-prey dynamics; bacteria (and plant) mediated nutrient and chemical cycling; and habitats that not only grow (and decline) over time, but that also serve as refuge for prey species. Expected changes in the Chesapeake system are simulated for 50 years into the future; marsh loss, submerged aquatic vegetation loss, TMDL attainment (for nitrogen and sediment loads), along with temperature increase are simulated separately and in combination, to estimate cumulative effects of these multiple factors, and to determine a dominant driver of change for the Chesapeake. Results from this work suggest that the water temperature increases expected for the Chesapeake system will be a very strong driver of productivity change, and that any work on other factors should consider expected temperature increases as well. Applications of CAM to specific Outcomes and Key Actions of the various CBP Workgroups to support the New Bay Agreement (2014) are also discussed.

Seminar Transcript

>> Thank you very much Scott and Phil. Thanks for having me here today. You know, as Scott mentioned, it is a model that's been available for a while now, but the reality is the Chesapeake Bay Program has not really been ready until this year with the development of workgroups and work plans and the outcomes. Now we've got the structure where we can build metrics that this model can apply and provide visualizations and simulations to look at the improvements that we're trying to make in the bay system. This is a full system model or and end-to-end model. It is set up as a full management strategy evaluation tool. It doesn't have to be used that way. Any particular module of it can be turned off. And that's actually important for the time it takes to run this model. If we're not using a portion of it, we definitely don't want to have that module on. And right now, the model is set up to mainly work on the left-hand of this -- side of this flow diagram. One of the most important things about this flow diagram is the double headed arrows. This is a modeling approach that we're attempting to model or account for the entire system. And so there's feedback throughout the modeling approach that are important to provide nonlinear dynamics. It's so important in an area like an estuary that we all work in. It's not a one-shot deal. There is a fairly large community of Atlantis modelers around the world. And we depend on each other quite a bit for model support and questions, problems we run into. The Chesapeake model is one of about ten of all of these spots that is actually working in an estuary. Most of the other spots are all on a shelf situation. As I mentioned, it's an end-to-end approach. So we're looking at all of these various factors and trying to account for them in the model and put the simulation of production, and that's what we're really after here. We're simulating production of all these various groups and it's in -- the production in the context of all these factors. So we're accounting for circulation, currents, temperature, and salinity with a ROMS model that underlies the Atlantis model that makes sure we stay in reasonable physical conditions that we're simulating. But that also allows us to directly manipulate things like climate change effects. So we can and we have simulated climate change effects of both salinity and temperature change. And the water clarity is another thing that the model accounts for specifically. So those water clarity affects have direct effects on things like SAV growth or microphytobenthos growth so this is the microscopic algae in the sediments that are so important in a benthic system like ours. And the habitats themselves, there's three habitats that we're simulating in the model. They don't just act as biological groups that grow or die off, but they also function as three-dimensional refuge for prey items. So then we can get at some of the importance of nursery areas of the bay. It is a brackish water model. I should point this out. It's chemistry that's hard-coded in the model is designed for marine systems. So we're not simulating up river of the brackish Allegheny portions of the bay. It does accommodate nutrient cycling throughout the system. And there's feedbacks there through bacterial action. There is a couple forms of bacteria as well as a wide range of detritus simulated in the system. So briefly, I'm going to give you a very overwhelming slide full of lots of words that you don't necessarily have to read or it's not going to be possible for you to read or digest, but I'm going to point a few things out that this model can support for the different workgroups in the bay program. As far as STAR is concerned, this is a very handy tool to actually test and develop ecological indicators. And it's a tool -- it's a great tool for integrating all of this data that we have. We've got an incredibly rich system full of long-term data accumulation here of 30 years or sometimes even more than that for this system. And that data is on all different scales. And what this tool does is once we've got all the data into the tool, we can easily look at differences in scale, differences in importance, of different factors. And it's a great tool to just integrate all the best available science that we have. In terms of water quality, we can directly visualize the improvements of the TMDL in terms of nitrogen and sediment loading. Because it doesn't handle the freshwater dynamics, it's not useful for phosphorus simulation. And, importantly, we can also -- this kind of it gets back at STAR. We can look at and demonstrate and quantify -- start to quantify the benefits of different monitoring approaches, different precision of monitoring, improving data gaps and simulate all these other outcomes, every one of these checkmarks that you're going to see on this list in terms of the TMDL. For climate, [inaudible] and I have worked together quite a bit looking at different applications of this model to climate questions. I'm not going to go through all these other checkmarks, but for the Habitat GIT we've had a lot of discussion there as well with Christine [inaudible] and some other members of the Habitat GIT and, of course, the Sustainable Fisheries GIT. This is the GIT we've developed under so they're very familiar with our efforts there but there's a lot of these. Every one of these checkmarks is a specific outcome or a key action that the model can be applied to. I should also point out that all of these things are not going to be addressed together. Each one of these checkmarks is probably a quarter's worth of work. There is only one person on my team working on this. So if we were able to actually identify more people that could work with me then we could start to address more of these checkpoints at the same time. But, again, very important, all of these checkmarks can be simulated in the context of TMDL improvements and climate change improvements or climate change effects, rather. So I just wanted to show you very briefly, some of the ways the output can be visualized. It's very challenging. Visualizing the output of this model, there is tons of output. Gigs and gigs of output comes from a model like this in the very simplest approaches sometimes up to terabytes of information. So every one of these boxes -- we'll see the structure of that in a minute, but every one of these boxes has a lot of information being cranked out of this model in any particular scenario. This is the fifth, actually a 20 year scenario. And I've just got it up here as an example of some -- of one way we can visualize outcomes. This particular example is a Bay Anchovy production on a yearly basis. And we're only looking at the production in this cell where there is a red dot. But you can have similar output for any of the boxes in the model. Every year, that flash of production -- we're looking at this fairly fast frame rate, but every year there is a lot of production of anchovy in every one of these cells. It is rapidly consumed but not to [inaudible]. These numbers are so very large that what we're seeing here is a rapid increase and rapid decrease. But at this point, for those on the phone I'll use a cursor, there is still plenty of animals in every one of these cells to continue production for the next year. I just wanted to give you an idea of how we can look at this data. We can look at this by tributary, for the whole bay if you want. You can also look at it by depth. At present, we're looking at an overview like an aerial view. But we could look at any particular layer and look at abundance by depth of any particular group or physical parameters that are output like [inaudible]. Another way of looking at output is in the point in time. This is a point 50 years in the future for striped bass populations on July 1st. Okay. And the comparison is between our current conditions and conditions under temperature increase and habitat loss of both SAV and marsh habitat. And they both -- figures in both panels share the color scale here. It is a little difficult to see on of a bay-wide scale. It changes. But the layered blue denotes more fish than the darker blues. Purple denotes essentially nothing is there. And what we're seeing is a little bit of a habitat squeeze. This is striped bass so they actually follow their prey when we've got hypoxic conditions developing in the middle of the summer. Now it is important that all the animals, all the groups, except for those like oysters that can't move, are allowed to move through the model to optimize their position in the system and where they're going to prefer to be because those are the best conditions for them physically as well as habitat-wise. A little bit about the design, it is a spatial model, as I said, so animals can move around. There is 97 odd shaped boxes to this system. And, again, those are shaped mainly for computational efficiency. We can't handle more than that really without really excessive runtimes. A current scenario for 50 years runs about two days. The cells are based on depth and salinity. Here they're color-coded by salinity just for your visualization of them. And in the main stem, we also separate out habitat that -- sediment that can -- oysters can recruit successfully too. So this is just sandy sand. It's not clay sand. It's not muddy sand, silty sand. This is all -- all of that is lumped into the mud boxes. So when we've got two boxes side-by-side, often times, one is the area and so these are specific areas. This is the amount of actual sandy sand available for oysters to recruit to and then the box next to it is actually the mud, area of mud. So that's why that was included. It's also a three-dimensional model. By depth, there's four depth layers at the most in the deepest portions of the bay. The first depth layer is for our photoactive layer. This is only a meter thick. And this is where the microphytobenthos is going to be growing for the most part on the edges as well as our SAV and emergent marsh grass. Then we go to the next layer is at five meters. That's the top of the [inaudible] line. It's assumed, this is a common assumption for the entire bay, so it's a very simplification. And then the bottom of the [inaudible] assumes to be at ten meters down. Anything deeper than that is a fourth water layer. And we set it -- simulate a meter of the sediment so we can allow things like bioturbators to have effects and groups like the clams to live in the sediment and crabs on top of the sediments. The rivers structure, you can't really see in that last slide. It's just too big of a scale but here's the James and the York. You see quite a bit of difference between how different TRIBS are simulated and that's due to their depth. What we have integrated in this box here is actually a [inaudible]. A good portion of those river tributaries to the York are included in that very little space. So although the model is accurate for volume of water, it's obviously not spatially resolved so we term it spatially representative of the bay itself, of the brackish [inaudible] of the bay. All TRIBS have this deeper layer through the TRIBS. And then the shallow layers on the edge. With the York, there is enough vertical distinction that we actually have more resolution here. This is a zero to two meters box, two to ten, and then the deeper channel whereas the James, this is a zero to 20 meter box. There is a lot of structure to our ecological groupings as well. There's 56 groups represented here. Obviously the predator groups that management cares about are included, but importantly so are all the forage groups that are thought to be essential. These are both vertebrate and invertebrate forage, medium-sized and small and protected species in those three habitat types that I talked about already. But then there's still a lot of holes that had to be filled in. We also have both groups of jellies, ctenophore, and sea nettles so we can capture the dynamics that are believed to be taking place between oysters and sea nettles and ctenophores. Those kinds of relationships we tried to capture as well with this structure. So very briefly, I want to illustrate the use of this model with an example that demonstrates the difference between modeling stressors singly versus modeling them together. So we can -- this is a model that we can start to get the cumulative effect of multiple stressors. Okay. So the stressors I'm talking about or factors in the system that are changing are habitat loss, both the emergent marsh and SAV loss, water column factors. We're talking about the TMDL essentially, attained levels of nitrogen and suspended solids and climate forcing -- we're talking about expected temperature increase. I also simulated salinity increase, but you're not going to see those results here. And I wanted to cover just those particular assumptions about each of those factors first. So we're talking about an assumption of a 50% marsh loss in the next 50 years. This is not arbitrary number. I know it sounds arbitrary, but this is a number that come up through our work with the [inaudible] habitat folks, and it's due to a combination of factors. Its shoreline armoring, subsidence that's taking place, and sea level rise, all are taking place concurrently in this system. And because once somebody does some hardening of their shoreline, whether it's riprap or bulkhead, the neighbors often do this too. So it's a somewhat arbitrary number but not entirely because it is believed that we could actually experience that degree of loss in this short time. And we're also simulating 50% loss of seagrass. And that is entirely arbitrary just so it's on the same scale as our marsh loss. We're simulating attainment of nitrogen loading improvements and sediment improvements in the bay, so a 25% reduction of nitrogen and the 20% reduction of suspended sediments. And our climate change assumptions are based on a stack workshop from a few years back that was published in 2010 that estimates a 1.5 degree increase of temperature. Now since then they've revised that estimate up. So now we believe the scale is more appropriately between two and three degrees Celsius in that time. So this is a conservative simulation. I also simulated salinity change and in that same report they estimated that the magnitude is going to be about two parts per thousand. It's going to vary during the year whether that's negative or positive. So I ran those simulations of both salinity decrease and a salinity increase in the spring. It's due to precipitation and essentially it was a wash. We're not seeing ecological change driven by this. And that's not a huge surprise because our groups are fairly well adapted to very highly variable environment. So I wasn't too surprised by that, but I'm not presenting those results either. I'm going to look at the results next. And what we're looking at is the status quo -- or this is the 2010 nutrient loading scenario here and everything in comparison to that. So these are current conditions for temperature, salinity, habitat, and whatever else I mentioned. These are the current conditions. And then we're going to look at the different scenarios on the y-axis. I'm going to have them simulated singly at first and then in combination with one another. This is just simply percentage change expressed on this plot from that single status quo scenario. And I'm only presenting a couple select groups. I could present, you know, 56 groups here. It's a little confusing. A little overwhelming. So I took a couple key groups that I selected ahead of time that would be of interest to show. Zooplankton, Bay Anchovy, which is actually one of the most important forage that we have in the bay. Another of the very important forage is the worm group which isn't just worms. It also includes other small invertebrates forage. And we have very good numbers for these, in fact, from our [inaudible] over the last 25 years. Blue Crab estimates, Atlantic Menhaden, Striped Bass. Some are flounder, another important fish predator but a benthic predator and then all fish, finfish averaged together. And when we look at that 50% marsh loss, we're seeing fairly small amount of variability here, about a 10% change from status quo overall. Bay Anchovy are on the bottom. There's always going to be winners and losers in every one of these scenarios. Whenever you're taking an ecosystem approach, there is going to be unexpected changes as well as those that readily make sense to you intuitively. Striped Bass, there's not much change at all. Zooplankton, not much change. Finfish, overall there is a decrease. In general, what we're seeing here is a loss of production for most of the groups, at least those groups I selected to show. When we looked at SAV change, much of the same picture. Again, about the same variability but importantly you're going to notice that the order of the winners and losers or the order of the losers, in this case, is very different between those two scenarios. When we look at the TMDL, now we're seeing that positive effects, more production with the TMBL conditions, but Atlantic Menhaden still on the losing side. But what you're going to see here is Atlantic Menhaden is always on the losing side in every one of these scenarios. And a lot of its life history happened outside the bay. And although the bay is important to its life history, it's never a winner in any of these scenarios. When we look at temperature increase, and this is simulated only -- the changes I simulated are only physiological improvements or increases, rather. So growth is increased. Clearance rates are increased. Movement is increased. And what we're seeing here is a much wider spread of the variability of the winners and losers. Now we're seeing very distinct winners and very distinct losers. Blue Crab is on our winning side. It is increasing its production fairly significantly, over 10% production increase. Not a big surprise. This is a southern species. And having warmer water actually is a better condition for Blue Crab. Our worms are doing well and benthic invertebrates that are acting as forage. Not so good are the rest of the groups. Bay Anchovy again is on the bottom. But, again, what I want to point out here is the order of the winners and losers. Very different. Very distinct for all four of these scenarios. So now we're going to start to do some combinations and look at accumulative effects. And here we're looking at temperature increase when the habitat loss so marsh loss and SAV loss. Again, very similar variability to the temperature scenario. And importantly, if you notice the order of the groupings is starting to stabilize here. And this time Blue Crab is on the winning side but not winning quite as much because they've got less SAV which is, as I've modeled them, their juveniles are dependent on that SAV. So that's a bad thing to have less of that stuff. Now we throw in everything together, temperature increase, marsh loss, SAV loss, and the TMDL improvements and, again, the similar scale that we see in all three of these last scenarios and Blue Crab again has increased benefiting from more SAV because of the clearer water. So generally, overall, we're seeing a dominant stressor of the temperature. There is other temperature affects that I haven't modeled yet. This is, again, just simply physiological change for all the groups. And other changes that could be modeled are the timing of life history events, the timing of migration in and out of the system, the timing of spawning events. And those are very important simulations to do as well because then we start to see matches or mismatches of the predator and the prey groups. And they're starting to be -- predators are starting to be exposed to prey that they've never eaten before because the prey weren't available or the other way as well. So in summary, oops, things are out of order here. The Chesapeake Atlantis model or CAM is a readily available tool to support a lot of the outcomes and visualize a lot of the things that the workgroups are trying to accomplish here. The temperature increase in our illustration has relatively strong effects compared to the other stressors or factors that are changing in the system. If you model the other stressors without temperature, you should expect that you're not -- you could get misleading information and that reasonable trends can be predicted if you happen to choose the right stressor to model and failing to do that can be a fairly significant problem. In this case, more than a 10% change of production of some of our most important predator fish. So one thing -- one of our next steps is looking at something else that was brought up earlier today in a couple discussions. One thing that I haven't included yet is acidification effects, and like temperature, that could have a major adjustment to some of the group's response to these other changes as well. So with that, just thanks to our funders and maybe we can have a little discussion or I can address some of your questions. [ Applause ]

Seminar Discussion

Coming Soon