Yellowstone: Ferocious Wild

Jennifer Sheldon, an independent ecologist, presents geodesign applications in ecology.

Jan 24th, 2013

Start From:
Player Color:

Right-click on these links to download and save this video.


00:01Today, I'm going to talk about geodesign applications in ecology...

00:04...using Yellowstone as a template for characterizing resilient and healthy ecosystems.

00:10I'm an ecologist with a background in systems approaches.

00:13Seventeen years ago, I cofounded a nonprofit based in Montana called Yellowstone Ecological Research Center...

00:19...where I served as senior research scientist, VP, and program manager.

00:27The greater Yellowstone ecosystem, about 20 million acres in extent, is a mosaic of public and private lands.

00:34In the center, you can see the park itself. It's about 60 by 60 miles in extent, a postage stamp, really, on the face of the earth.

00:42In Yellowstone, we have an ecosystem whose structures - protected undeveloped landscapes...

00:48...components - that is, species - and processes, for example, migration, predation, and hydrological regimes...

00:55...are among the most intact remaining in the temperate world.

00:59The "ferocious wild" of my title refers to the fact that all 18 species of terrestrial carnivores are originally endemic to the park...

01:07...or still present, one of the few places in the lower 48 where you can still see a wolf, a mountain lion, a grizzly bear roaming freely.

01:17Given this happy conjunction of ecological integrity and legal protection, Yellowstone offers itself as a benchmark...

01:24...or standard of reference, for the healthy, well normal ecosystem.

01:29Using Yellowstone as a model, the idea is to build a library, a set of reference standards, for healthy, functional ecosystems.

01:38The larger idea is to develop methodology and workflows for establishing reference values to characterize healthy ecosystems in general.

01:48Over the last six years, I've worked on a joint NASA/US Fish and Wildlife Service project...

01:53...developing tools, workflows, and imagery to build holistic ecosystem models.

01:58If any of you are interested in the more technical aspects of this project, find me later...

02:03...but for now, in a general sense, when I use the term benchmark ecosystem, what I'm getting at is this...

02:12...a systems approach to reference standards.

02:18Of course, all this begs the question, "What constitutes health?"

02:23Take a look at the two people in the image. Are they healthy?

02:27How would we know? What are the markers that we use?

02:30Then take a look at the setting - a lake, looks alpine, water looks clear, air looks clean, landscape appears undeveloped.

02:39That's often a kind of shorthand for ecosystem health.

02:42There are a lot of implicit assumptions we make based on outward appearances, but we know intellectually...

02:48...that appearances provide insufficient information about vulnerability, about resilience, about long-term health prospects.

02:56The look is not the metric; the look can't be the metric of health when we deal with ecosystems.

03:02We're left without adequate metrics. So what do we do?

03:13It strikes me that the most reasonable way that we can all approach ecosystem issues is to borrow from public health paradigms...

03:20...where it's established practice to operationally define health by both function and range of values.

03:28Think of your blood panel when you go to the doctor. How do you know you're sick?

03:32You've deviated measurably and directionally from a standard of published, validated, well normal conditions.

03:38These aren't political; these are reference standards.

03:42There's a library of ranges of values for almost all aspects of individual and population health.

03:49Think of infant mortality rates, influenza infection rates, your white blood cell count, your serum cholesterol.

03:56We're all quite comfortable with this system of metrics. Why not apply the same principles to ecosystem health metrics?

04:04We sort of do through the Clean Air and Clean Water acts, but that's as far as we've gotten.

04:10My work is about creating and supporting the library of well normal values for intact ecosystems.

04:20This approach moves the arguments about health and resilience into the empirical, data-driven domain.

04:26We remove the frame of contentious philosophy and move ahead with rationality informed by long-term vision of systems persistence.

04:35Ecology's historically been confused about its ability to deliver a clear outcomes-driven message.

04:42Let's contrast this with the thought system as it exists in the practice of medicine.

04:46In medicine, a code of bioethics informs all activities, whether you're a basic scientist, a bench researcher, a clinician, a family physician.

04:56All practitioners share an outcomes-driven, common code.

05:01In ecology, we experience a continuous recalibration of what we accept as minimal ecological integrity...

05:08...when we don't have reference standards or a guiding code of bioethics to go back to.

05:13But we can do better.

05:16Here's a workflow for the ecological benchmarking effort I've been working on...

05:21...supported in part by a NASA ecological forecasting grant... integrate remotely sensed data, organismic data, and geophysical data.

05:29We use geodesign principles throughout in workflows and thought process...

05:33...and in framing applications and approaches to complex ecological data integration.

05:39The fundamental workflow in all cases is as shown.

05:42So we begin with species and/or habitat data. These are available either present day or retrospective.

05:49We add in, from the huge armamentarium of available climate inputs, both retrospective and prospective.

05:58Yellowstone's one of the most intensively imaged areas globally, both because it's of interest ecologically and also because it serves... an analog for remote theaters of interest - highly dissected mountainous terrain, low population densities, low road densities.

06:17So lidar, small-footprint lidar, hyperspectral, radar, aerial photography - you name it - and there's a number of group shoots as well.

06:31These first three inputs are then fed into existing, already validated, well-proven ecological model structures...

06:37...and from those, we can take out support for management and decisions.

06:45This work focuses on animal populations because animals are aggregators of ecosystem condition information.

06:52The better the habitat quality, the more the animal community organization will be stable, functional, and persistent.

06:59Our modeling approaches make use of this information aggregation.

07:02We focus on animal densities as a highly efficient shorthand for ecological integrity.

07:08The converse is, of course, also true.

07:10Absence of animal populations shows us that a piece of the system is missing or out of the range of sustainable values.

07:24I'm going to quickly show you some work from a systems model in Yellowstone.

07:28What you're looking at on the left are the building blocks of a holistic ecological model, spatially specific data arrayed in layer stacks.

07:35Think of these as maps of attributes of interest.

07:38On the left side of the image, we begin with a USGS quad orienting you to the region of interest.

07:44Below that, there's a small mammal biomass layer; basically, how many grams of small mammal per square meter.

07:50Think of small mammals as the krill of the terrestrial landscape, critically essential for the food web.

07:57Below that, a remotely sensed derived vegetation layer...

08:00...and elevation is, as you all know, critically important for prediction of response to changes in climatic regimes.

08:10Moving across the image, there's a little box called Data Integration.

08:15For those of you who live in the GIS world, that box can also be called Blood, Sweat, Tears, and Teeth Gnashing, right?

08:23A lot happens in there, so that's about 90 percent of the work in assembling these models occurs in that component.

08:31And then we move across to the right-hand side of the model, which is the layer stack array. And let's see.

08:42So we can see the same data layers connected by a punch-through, that purple rod, moving through all the data layers.

08:47The intersection of this punch-through with each of the data layers yields a value from the point of intersection...

08:53...and these are subsequently fed into tabular arrays, spreadsheets, for the follow-through into the modeling component of the workflow.

09:03Obviously, these purple punch-throughs are arrayed across a landscape...

09:07...and they can be developed according to different sampling regimes randomly or different sampling intensities.

09:13So the appeal of this layer stack and then punch-through approach is threefold.

09:18It's absolutely simple; it's scalable up or down on several levels; and lastly...'s amenable to scenario building or what-if models, as we'll see in a moment.

09:28So let's just take a look at the right-hand side of this image in an expanded view.

09:34Same punch-through, here shown in red so it's a little bit more visible for you guys.

09:38On the layer stack of the data are fed - the data intersection values are fed into a tiny fragment of a table...

09:46...that you can see up there as a representative grab, and then again, these go into the models.

09:52So you all remember the concept of the well normal ranges of values for healthy ecosystem that we began with.

10:00This table can be immediately mined for high and low values across the region of interest, so we have a bankable range of values...

10:07...for each of the attribute layers, as well as a wealth of relational information and distributional information.

10:14For scenario building, data layers for anticipated futures, such as proposed developments like new roads, can be input...

10:21...and the resulting forecast models can be compared to existing models.

10:25We hold all inputs constant, save one, rerun the model.

10:28So this is a really controlled, rigorous, and transparent way of forecasting.

10:33The bottom layer in the layer stack is the model output itself.

10:37In this case, it's a probability surface for an antelope habitat model, and you'll see it again in the next image.

10:42If there's any modelers out there, it's a resource selection probability function map.

10:48And so here we see the modeled output.

10:51On the colored image at the top right represents about 15 hundred square miles of Yellowstone's northern range.

10:57You can see the USGS quad that orients you to this piece of landscape...

11:01...peeking out from underneath the same-scale color ramp modeled surface.

11:05Red line running through the center is a road.

11:08And as far as the color-ramped image itself, this image answers a question - Where's it great to be an antelope?...

11:14...with brown representing prime habitat and blue representing less used or less desirable habitat.

11:23So the color ramp surface here is arrived at by using equations that relate all environmental layers to each other...

11:30...then we fit the best model from the arsenal of current published ecological model structures.

11:35The table to the left side of the image depicts that part of the quantitative model results.

11:39I figured since this is a black hole of postlunch slump, we can leave that part out, right?

11:46So let's take a quick look at the lower right corner of this color-ramped image, where I've thrown in a yellow box, and we'll run a scenario.

11:54Here we see that grab of our original model surface.

11:57I have run a road through there in tan; it's evident and circled in yellow so you can see it.

12:03It's clear that that road runs through what we remember is prime antelope habitat in dark brown.

12:09So remember, for this process we hold all model inputs constant, save the one, and then we rerun the model.

12:16And we can see in the bottom image this rerun output.

12:19The total coverage of brown pixels by visual inspection is decreased...

12:24...and the road effect is also more pronounced at the convergence of the two roads.

12:29So here we have a quantifiable, transparent, and defensible forecast model.

12:39Remember that animals are information aggregators. They tell us directly about ecosystem quality.

12:46So we build models around animals, and their continued population presence is the most efficient way to understand ecosystem integrity.

12:55Animals provide the fundamental controls on the systems they inhabit.

12:59Our dominant cultural narrative portrays animals as decorations on the landscape...

13:05...but recent research shows us how integral they are to knitting together the fabric of the systems we rely on.

13:14Yellowstone is the planet's first national park, protected in perpetuity. Not just 250 years; perpetuity.

13:23That's a bold concept Roosevelt came up with.

13:26Her legacy has inspired the creation of parks and protected areas throughout the world.

13:31With all the components and processes still in place, she and the data that she embodies...

13:37...can be used as a standard of reference for system health - the world's first benchmark ecosystem.

13:47In conclusion, I agree with Bran Ferren's assertion that the stories we tell about the world we live in matter greatly.

13:54I believe in the power of narrative models and of visual and graphic imagery.

14:00These have the power to shift how we humans relate to the world.

14:04I think we can begin to write a story that's large enough to encompass a bioethical set of arguments.

14:11The story I'd like to write is a story of a data ark, A-R-K, in which to place our knowledge of the healthiest of what remains... that when we design our way forward, we still maintain the blueprints of the elements...

14:28...that we have not yet become wise enough to understand, let alone replicate.

14:35So I hope you take away three ideas.

14:37First, without a standard of reference, our conservation efforts will necessarily be incomplete.

14:43Let's benchmark our conservation efforts using state-of-the-art data.

14:47Second, animals are brilliant aggregators of information about ecosystem conditions.

14:52We can use their presence to inform us on how we're doing in supporting earth's systems.

14:59And lastly, a complex effort like this needs a superstructure, a powerful yet transparent architecture...

15:07...and geodesign can serve that function.

Copyright 2016 Esri
Auto Scroll (on)Enable or disable the automatic scrolling of the transcript text when the video is playing. You can save this option if you login


No comments. Be the first to write one below.

Comment on this Video