Drought Refugia: Remote Sensing Approaches and Management Applications

Video Transcript
Download Video
Right-click and save to download

Detailed Description

This webinar was recorded on June 12, 2017 as part of the Climate Change Science and Management Webinar Series, held in partnership by the USGS National Climate Change and Wildlife Science Center and the USFWS National Conservation Training Center. Webinar Summary: During droughts, localized areas of the landscape (drought refugia) retain surface water and soil moisture needed to sustain wildlife and vegetation. Remote sensing from satellite imagery offers powerful tools to identify refugia and study their responses to changing weather patterns over time. This talk will present two recent applications of remote-sensing analysis related to drought refugia research in southern Oregon. In one study, hydrologic resilience of springs was inferred using remote sensing of groundwater-dependent vegetation in a semi-arid sage steppe ecosystem. In another, refugia from drought and mountain pine beetle were identified in lodgepole pine and whitebark pine forest. Efforts are currently underway to scale up identification of drought refugia to the Pacific Northwest region. This talk will also discuss the integration of natural resource management priorities with refugia research to provide managers with new information to support ecological monitoring, restoration, and conservation.


Date Taken:

Length: 00:52:39

Location Taken: US

Video Credits

Jennifer Cartwright, USGS Lower Mississippi-Gulf Water Science Center;
John Ossanna, FWS National Conservation Training Center;
Elda Varela Minder, Holly Padgett, Shawn Carter, USGS National Climate Change and Wildlife Science Center


Shawn Carter:  It's my pleasure to have Jen here with us today. Jennifer Cartwright's an ecologist and geographer with the Tennessee office of the Lower Miss‑Gulf Water Science Center. She's broadly interested in understanding the drivers of biodiversity across a range of ecosystems, in particular ecohydrology, which is the study of how water interacts and affects natural communities.

Her previous work has examined how water management practices affect fish communities in streams, and how climate change may affect island‑like ecosystems in the southeastern U.S. Her current work focuses on drought and climate change in the Pacific Northwest. Jen's here to talk about some of her work, and it’s my pleasure to introduce her. Here she is.

Jennifer Cartwright:  Hi, everybody. Can you hear me OK?

John Ossanna:  We're good.

Jennifer:  All right, good. Thank you everyone for being in this webinar. I want to thank the National Climate Change and Wildlife Science Center for inviting me to speak, and I want to thank the Northwest Climate Science Center for supporting this work.

My email address is on the screen. We will have some time at the end for questions and discussion. Also, if you have any other ideas, or suggestions, or questions for me, please don't hesitate to contact me.

I'm going to start by explaining what we mean by drought refugia, or hydrologic refugia. Then I'll present some preliminary results from a couple of studies in Southern Oregon, using remote sensing to look for refugia, and I'll briefly touch on some work to scale up some of this analysis to the regional level.

Finally, I'll talk about some ongoing, collaborative work to really connect the science around refugia to the needs and priorities of natural resource managers.

To start, I want to highlight two recent papers that are really helpful for anyone that would like more background on these concepts. The first is by Morelli et al., it came out last year in "PLOS ONE," called "Managing Climate Change Refugia for Climate Adaptation."

The second is by McLaughlin et al., it came out earlier this year in "Global Change Biology," called "Hydrologic Refugia, Plants, and Climate Change." These are two excellent references if you'd like to get some more in‑depth background from these two review papers, on these concepts of refugia.

This figure is from Morelli et al., 2016, and it's just showing a variety of landscape features that might promote climate change refugia. Basically, the idea is that as climate change is progressing – and that could include warming, in some cases, in some locations, it might include climatic drying and drought intensification - that those effects are not going to be homogenous across the landscape.

There may be localized areas that are somewhat buffered, where overall change is less, or perhaps is changing less rapidly than the surrounding landscape. There's been a fair amount of interest in climate change refugia over the past decade or so. A lot of that work has focused on temperature buffering from warming and somewhat less emphasis on hydrology and soil moisture.

In places where climate change is expected to alter the duration or the intensity or the frequency of drought, we can imagine that drought refugia, areas of landscape that retain biologically available water during drought could become increasingly important to wildlife and vegetation.

This figure is from McLaughlin et al., 2017. I find it really helpful to remember that refugia are not necessarily places where nothing is changing. This figure conceptualizes three kinds of refugia, and this is in the context of climatic drying, possibly drought intensification.

To the left, you have a stable refugium. This could be a place, hydrologically, where things are not changing through time with climate change, whereas to the right, you have a relative refugium. Even within this refugium, hydrologic conditions could be changing, in this case, getting drier.

The hydrologic requirements shift from species one to species two, you have the ecological and hydrologic change happening even within the refugium, although it remains moister than much of the surrounding landscape.

On the right, you have this idea of a transient refugium, which is where if there are certain climatic or climate change thresholds that get surpassed, then that landscape location just might not be able to fulfill that function at all anymore.

How would we go about finding these locations on the landscapes, for example, refugia during drought? There's no question that field observations and measurements on the ground during droughts are critically important. For one thing, we need to understand species‑specific responses, and we need to understand interspecies interactions that are helping to define refugia.

Droughts can affect vast landscapes. In the case of remote areas and rugged terrain getting field observations can be difficult and expensive, and it's almost impossible to sample the entire landscape.

At this point, we have these hypotheses about where we think drought refugia or hydrologic refugia might occur on the landscape. How do we go about testing these hypotheses? Remote sensing offers some powerful tools in this.

In the case of the Landsat archive, we have data that goes back decades, every 16 days at 30‑meter resolution. That allows us to look back at past droughts, and it also allows us to look more generally at variability in climate and hydrologic conditions through time, and see if we can find vegetation signatures in the remote sensing data that can help make some inferences about hydrology.

If we can find areas on the landscape that seem to be more resilient than their surroundings in terms of how they respond to droughts or interannual variability and water availability, then those are candidates for these kinds of refugia.

Now I've given you a brief background. I'm going to jump into some preliminary results from the first of two studies in Southern Oregon. This first one is asking whether springs, groundwater fed springs, could be refugia in a changing climate.

I'd like to acknowledge my collaborator in this, Hank Johnson, also with USGS at the Oregon Water Science Center. In arid and semi‑arid landscapes, springs can be important sources of surface water and soil moisture for wildlife and vegetation.

Springs have been mentioned as climate change refugia. In fact, both of those review papers that I showed you at the beginning, both of those specifically mentioned springs as possible refugia. It may be somewhat complicated because we know that springs even in close geographic proximity to each other can have very different hydrologic responses to climate drivers.

Springs that are dependent on recent precipitation and snow melt for their recharge might actually be pretty vulnerable to drought intensification and, in fact, some springs might run dry all together. That could be an example of that transient refugia idea after a certain threshold is surpassed.

In general, we would expect that springs that are discharging older, deeper groundwater might be more resilient to climate change in the form of drought intensification. Those would probably have the greatest likelihood of serving as hydrologic refugia.

The problem is we don't have a lot of hydrologic data for many of these springs. For example, this study area is the Steens Mountain Wilderness in southeastern Oregon managed by the Bureau of Land Management, BLM.

There are hundreds of springs mapped in the national hydrography dataset in this area, but for the most part they are just points on a map. Most of them have a few if any hydrologic records. As a result, we generally don't know the mean annual flow of most of these springs, much less do we know how that flow is changing overtime in response to droughts or climate change.

If we zoom in on individual clusters of springs, and we look at high resolution satellite imagery, we often see these characteristic green areas surrounding a down gradient of springs. That led us to ask the question of whether we could use remote sensing of the vegetation to make some inferences about hydrology and start asking some of these questions about resilience.

We used Landsat data from 1985 to 2011 to do two things. First, to delineate these green areas surrounding springs, called surface moisture zones, or SMZs. Second, to follow them through time and look at how they respond to climate drivers.

July normalized difference vegetation index, NDVI, is a spectral index from remote sensing. Basically you can think of it as vegetation greenness.

We developed seven indicators of resilience and follow it based on this time series analysis. Then we synthesized all those into an overall resilience indicator and tried to situate groups of springs along a gradient from most to least resilient within this study area.

The first two indicators were just the mean and the standard deviation of NDVI through time. The most resilient SMZs, these green areas around springs, in blue, generally had higher NDVI that was more stable through time, compared to the least resilient groups of springs.

We also wanted to look at the difference in greenness between the SMZs, these green areas around springs, and the surrounding landscape.

For example, on the left, you see a case of a resilient SMZ, where the greenness in that area was quite a lot higher than the greenness of the surrounding landscape. On the right, you see a less resilient cluster of springs, where the NDVI is still higher. That area is still greener than the surroundings, but that difference is less, and it's less stable through time.

In this area, topographic controls on precipitation and snowpack are really important. We did some simple linear regressions of July NDVI and the 90‑day antecedent precipitation. That would be the April to June, that early growing‑season precipitation. Then we also did simple linear regressions of July NDVI with the previous winter snowpack.

Here, for example, on the left, you can see a case where that July greenness was relatively decoupled from the 90‑day antecedent precipitation. Whereas on the right, you can see there's a stronger, tighter association between that greenness and antecedent precipitation. That can also give us some information about possible resilience to drought.

Finally, we took a five‑year period as an example of interannual climatic drying in the early growing season using a year that was exceptionally wet, 1998, in terms of that 90‑day antecedent precipitation, followed by four years that were drier than average, and looked at how those groups of springs responded over that time.

The most resilient SMZs declined somewhat, but that response was much less intense than for the least resilient SMZs that saw a greater decline during that time of interannual drying of those early growing‑season conditions.

Putting all these seven resilience indicators together, we saw that they were all highly intercorrelated, so we used principal components analysis to derive an overall resilience score, and then looked at how that related to certain characteristics of the landscape.

For example, the resilience scores were positively associated with the mean elevation and mean slope. In this study area, we are seeing that the groups of springs higher up on the mountain and more steeply sloping topographic positions were showing greater resilience.

The resilience scores were not correlated with the number of springs in each cluster, which might sound surprising. But remember, we still don't know the hydrology, or the flow, of any given spring, so it's not always going to be the case that more springs necessarily equals more surface water on the landscape.

The resilience scores were also not associated with the size of the SMZ. That means that green areas on the landscape that were larger were not necessarily more resilient through time.

At this point, what does this analysis tell us, and what are we still trying to work through, what questions do we have?

The big advantage of using remote sensing for these kinds of springs is that we can look at relatively large numbers of springs in areas that are more challenging to do it in the field. Of course, we have to integrate and calibrate this remote sensing with field observations.

At this point, we have some preliminary assessments of how springs have responded to climate drivers in the past that may give us some information about how they might behave in the future. But we need to integrate that and calibrate it with some field observations.

The USGS Oregon Water Science Center is hoping to embark on a multiyear effort looking at the hydrogeology of springs in southeastern Oregon through field observations. The Nature Conservancy in Oregon has done a lot of work on groundwater‑dependent ecosystems, including a lot of field observations of flow in more forested springs.

Those are the kinds of observations that need to be integrated with this remote sensing approach if we're going to get a handle on predicting the possible hydrologic and subsequently ecologic resilience of springs to climate change, and possibly be able to identify possible refugia.

There's some management applications of these questions because a lot of springs have been altered ‑‑ for example, diverted to provide water to livestock ‑‑ and there's some interest in restoring springs so they can serve more of their ecological function.

Of course, there's limited conservation dollars for doing that, to the degree that we can provide information that's useful to natural resource managers, that may help them in some of their prioritizing some of their efforts of restoration and long‑term monitoring.

Now we are going to switch gears, and we're going to look at a very different ecosystem, but also one in which water availability is critically important. I'll present some preliminary results looking at refugia from drought and mountain pine beetle in a whitebark and lodgepole pine ecosystem.

In forests across western North America, drought and insect outbreaks are primary disturbance processes linking climate change to tree mortality.

We can use some of our knowledge about the physiological effects of drought and insect outbreaks, and combine that with our understanding of the landscape controls on microenvironmental conditions to try to generate some hypotheses about where we might find refugia from these kinds of disturbance.

The main objective of this study was to evaluate some of those hypotheses, first, by identifying refugia using remote sensing approaches and then, by modeling the landscape controls on the locations of refugia. I'll start by giving you an overview of the methods. Then I'll walk through some of those hypotheses and the evidence that was related to each one.

The study area is the Gearhart Mountain Wilderness in the Fremont National Forest in southern Oregon. It's about 120 kilometers east of the Cascades in a dry conifer forest transition zone between the wetter Cascades to the west and the semi‑arid sage steppe of the Great Basin to the east.

In the low elevation valleys of the study area, it's mostly fir and ponderosa pine and some lodgepole pine. At around 2,100 meters, there's an ecotone to more homogeneous lodgepole stands. Above about 2,200 meters, there's whitebark pine and mixed whitebark/lodgepole stands.

The disturbance history of the study area includes a couple droughts, one, a single‑year drought in 2001, and then a multiyear drought from 2007 to 2010. Two years were selected for identification of refugia based on an integrated assessment of the disturbance history in the study area. That was 2001 and 2009.

Mountain pine beetle is endemic in this area, but there were no major outbreaks of mountain pine beetle from around 1980 to 2005. Starting around 2006, we see the beginning of this really severe and widespread mountain pine beetle outbreak that affected over 80 percent of the study area at its peak, around 2007 and 2008.

This photo was taken in 2012, so it gives you an idea of the extent of the mortality in the area. The study area, I think, is well suited for identifying refugia from drought and insect outbreaks, in part because we have a couple different droughts to look at that differ in their magnitudes and their disturbance interactions.

This study area hasn't had any recent fire, and because it's a designated wilderness area, it's also protected from other disturbances, like land use change or logging, that might otherwise make it more challenging to isolate and identify the effects of drought and insect outbreaks.

These are plots of normalized difference moisture index from Landsat in August of each year from 1985 to 2011. NDMI has been shown to be a good indicator of canopy water content and forest responses to disturbance. You can see that from around 1985 to about 2000, which are the shaded boxes in each plot, the NDMI was fairly stable for each of the forest types.

Then in 2001, which is that first dashed line, there's a slight decrease in NDMI in each of the forest types. That was that first single‑year drought that had no insect outbreak. For the most part, NDMI rebounded after that.

Around 2007 or so, you start to see these dramatic declines in NDMI, especially in the lodgepole and whitebark stands. That was the onset of that multiyear drought that coincided with the mountain pine beetle outbreak.

Based on what we know about how mountain pine beetle outbreaks occur and progress, it seems likely that the drought played at least some role in triggering that outbreak. They're not necessarily independent of each other.

Just from the Landsat data, we really can't know ‑‑ say, in 2009, which is that second dashed line ‑‑ which of those effects are from the drought and which are from the insect outbreak, because within any given Landsat 30‑meter pixel, those effects are integrated.

I would argue that drought and insect outbreaks in forests across western North America form a disturbance regime or a disturbance syndrome acting often together. From a land management perspective, it might be useful to look for refugia from their combined effects, which is what the Landsat data allows us to do.

The first step in identifying refugia was to calculate a drought year anomaly. For each of those years, 2001 and 2009, that was done with a simple difference from a reference NDMI, which is just the median value for each cell, for each Landsat pixel from 1985 to 2000, which was that period of relative stability that I showed you.

In the bottom two boxes, you can see that the anomaly values in 2009 were a lot more severe than the ones in 2001. Here are those anomalies broken out by forest type. We can see that in 2009, pretty severe anomalies, especially in the lodgepole and the whitebark stands, which is what we would expect during a mountain pine beetle outbreak.

In 2001, we see these more moderate anomalies across the board. That was that single‑year drought with no insect outbreak.

To define and identify refugia from this, there's a number of ways that could be done. A simple approach is just to take some percentile value of the highest, meaning the least negative, anomaly in a given year.

Here's an example in lodgepole pine and whitebark pine in 2009, if we say that we want to designate the top 10 percent of each of those forest type areas as refugia based on having the least severe anomalies in those years.

Now I want to start talking about some of the hypotheses about where these refugia might be located, why they would be there, and what we found. Boosted regression trees, which is a machine‑learning approach, was used to model the landscape controls on refugia locations.

I'll walk through some of these hypotheses.

The first one was elevation. The hypothesis was that we would see more refugia at higher elevations. That could be because there's greater precipitation, maybe less evaporative demand at higher elevations, possible to have a greater snowpack that persists longer into the growing season. In this study area, using these methods, we did find support for that hypothesis in the lodgepole stands.

The vertical axis here is the probability, at any given elevation, of finding refugia. We see more refugia in lodgepole stands at higher elevations. But in the whitebark, in the lodgepole/whitebark mixed stands, we see the opposite. There's more refugia at lower elevations. That might be related to some of the topographic characteristics of some of those lower elevation sites that I'll show you in a minute.

For slopes, there was no a priori hypothesis about the relationship, in part because previous studies of mountain pine beetle impacts have found positive relationships to slope, or negative or no relationship to slope.

On the one hand, steeper slopes promote faster runoff, possibly less infiltration of rainfall and snowmelt, so possibly drier soils. On the other hand, steep slopes that are leeward could accumulate a deeper snowpack. If they're topographically shaded north‑facing slopes, they might have snowpack that persists later into the growing season.

Using these methods in this study area, we found more refugia on steeper slopes for each of the forest types. Related to that, we found more refugia on topographically shaded slopes.

Topographic heat load index integrates slope aspect and latitude to give this overall metric of solar insulation or topographic shading. The hypothesis was to see more refugia on shaded slopes, which we did see in each of the forest types.

Shaded slopes, possibly less evaporative demand, possibly greater snowpack that persists longer into the growing season. We know from previous studies that there's greater soil water retention on those shaded north‑facing slopes. We did find support for that hypothesis.

Topographic position index gives an indicator of how convex or concave a landscape position is. The high values of TPI are those convex locations, like ridgetops. The negative TPI values are those concave locations, like valleys.

The hypothesis is to see more refugia in the low‑TPI areas, like valleys, possibly because of cold‑air pools. Cold‑air pools are these temperature inversions that can happen in mountainous terrain. They can help promote dew formation and retain some humidity, possibly helping to ameliorate some of those effects from drought.

In the whitebark stands, we do see more refugia in those convergent valleys. That might help explain why we saw more refugia at lower elevations in whitebark down in those valleys. Whereas with lodgepoles, we're seeing more refugia along these convex ridgetop locations. That's looking like these topographically shaded high‑elevation ridgelines, in lodgepole.

We could also imagine that soil characteristics could play some role in where refugia might be located. The hypothesis with soil bulk density was to see more refugia in less dense soil. That could be because less dense soil promotes greater root elongation and proliferation. We did, in lodgepole, see more refugia in less dense soil. Not as much in the whitebark stands.

The last one I'll show you here is that forest stand characteristics also might play a role in where refugia could be located. For example, we had the hypothesis of more refugia in thinner forest stands, for a couple reasons. Just from a drought perspective, thinner forest stands could have reduced competition in the form of transpiration for those limited soil water resources.

From the perspective of mountain pine beetle vulnerability, there's been some evidence that thinner stands are less vulnerable, for example, because of greater wind penetration can help disperse some of the pheromones that mountain pine beetles use to coordinate their attacks.

Here, we did see especially in lodgepole, more refugia at these lower stand densities. Those are some of the modeled relationships. I thought it could be helpful to get even a little more grounded on the landscape and maybe look at some examples of the kinds of landforms that might be associated with refugia.

I'll show you three examples. These are these green boxes, and we'll start at the north, and then we'll work our way south. This is an example of a location that was identified by the remote sensing as being a refugium. It's a steep, north‑facing slope.

This is also a steep, north‑facing slope. It's at the head of a cirque, which is a glacial valley, and the aerial image here has been made from 2016, so this is a late June aerial image. You can see there's still some residual snow, even at that time of the year, uphill of where the refugium was identified.

It's interesting to consider whether some of these snowmelt timing dynamics might have some role to play in hydrologic refugia and where they might be located. Then, this last example is one of those convergent environments.

The blue line is a stream segment from the national hydrography dataset, and the refugium that was identified in the remote sensing is adjacent to that riparian area, downslope from a steep, northeast‑facing slope.

Again ‑‑ as with the study on springs that I showed you earlier ‑‑ we are going to have to integrate field observations, during droughts, during disturbance events, with the remote sensing if we're going to get a solid handle on where refugia are located and the kinds of landscape and hydro‑ecological processes that support refugia.

Now, I want to zoom way, way out. From a particular landform in a particular study area, now, let's just zoom out to the entire Pacific Northwest, including Washington, Oregon, and Idaho, to ask whether we could identify refugia at that scale. I want to acknowledge my collaborators in this, Josh Lawler and Julia Michalak at the University of Washington and Solomon Dobrowski at the University of Montana.

Trying to identify drought refugia at a regional scale is a big proposition, in part just because this region is so diverse. You've got everything from temperate rainforests on the Olympic Peninsula to near deserts in the Great Basin.

Here again, we do have remote sensing. This data that you see is anomaly in enhanced vegetation index from MODIS satellite imagery at about one‑kilometer resolution in the summer of 2015, which was a drought year in some parts of the region.

The areas that have missing data are where we masked it out, because the landscape doesn't represent natural vegetation, either because it's developed or agricultural or else because there's been recent disturbance in the form of fire or logging, that kind of thing. Even at a regional scale, we do have hypotheses for the kinds of landscape features that might support drought refugia.

That could be things like shallow groundwater, prevalence of springs, valley bottoms that might support those cold air pools I mentioned, topographic wetness, wetlands and riparian areas, and some of those north‑facing shaded slopes.

Doing it at this scale, there's a number of challenges. One of them is that the landscape is such a mosaic, a patchwork of natural vegetation and human‑altered landscapes and land uses.

Another challenge is that it's very possible that the ecohydrologic processes that might support drought refugia, say, in a high‑elevation forest, could be very different than those that are operating maybe in a lower‑elevation sage steppe, for example.

These are some of the challenges that we're going to be wrestling through over the next few months, so stay tuned, but in the rest of the time I have, I want to get at this, "So what?" question. Why do we care about finding refugia at all?

In particular, how can we relate this kind of science to the needs and the priorities of natural resource managers? For this, I want to highlight the work of Aaron Ramirez with NCEAS. He is the coordinator of the Refugia Research Coalition, the RRC, in the Pacific Northwest that is supported by the Northwest Climate Science Center. His email address is on the screen.

If you would like more information or to get involved in that effort, please contact him. I will do my best to represent the work of this group that I'm honored to be a part of, but he's the coordinator, so he could probably get you connected and answer questions if I'm not able to.

The Refugia Research Coalition in the Northwest is a group of about half scientists and half natural resource managers. One of the first things that we did upon forming as a group was to do an informal survey of natural resource managers and ask a number of questions.

One of them was, "In what ways could research on refugia help address your priorities as a natural resource manager?" I'll review a few quotes that we got. One manager said, "If there are habitat components and associated species that will be maintained because of climate resiliencies, then these would become management priorities."

Another said that, "Refugia science would help us prioritize where best to invest our limited conservation and restoration funding." Another said, "The ability to identify areas that are likely to serve as refugia would be useful in prioritizing potential property acquisitions."

This common thread when talking to natural resource managers is that the science around these various kinds of refugia could be useful in helping them prioritize, because they have a lot of decisions they have to make on the landscape. There's a lot of needs, and they have limited resources for doing all of that.

Now, I don't think anybody would claim that identifying refugia is the only or even the primary consideration in making a conservation and management decision, but it's one tool in a larger toolbox of climate adaptation. Everything that the Refugia Research Coalition in the Northwest has been doing is situated within this climate change refugia conservation cycle.

These ideas and this figure are from Morelli et al., 2016, one of those papers that I showed you at the beginning. It shows that just identifying where refugia are, that's just one step in this larger process that is iterative and cyclical and involves close collaboration between scientists and managers.

Some of the early work of the RRC has been on some of those initial phases like defining the planning purpose and objectives, assessing climate impacts and vulnerabilities, and looking at those conservation goals and objectives. This is the overall approach, and it starts with the management priorities and clarifying the decision‑making context.

The idea is that natural resource managers have to make a lot of decisions in their work, and some of those decisions may be well‑informed by the existing science. In those cases, synthesizing the science that exists can help support and give information to managers to help make some of those decisions.

In other cases, there may be places where the scientific knowledge just isn't there yet, or it's still emerging, so we can identify some of those knowledge gaps and try to identify science needs for future research. The goals and process of this group, to start with, clarifying the decision‑making context and identifying management priority areas.

When we had our first in‑person meeting as a group, we, as a group, identified four broad ecosystem types that were going to be priorities in the context of the Pacific Northwest. Those were late successional forest, coastal and estuary systems, sage steppe and sage grass habitat, and riparian and river systems.

You can imagine how within those broad categories, you could drill down and identify within those, a number of management priorities. For example, within late successional forest, you could have a number of priorities within that that might include, for example, some of those high‑elevation five‑needle pine, like the whitebark refugia idea that I was showing you earlier.

The ongoing work is to synthesize the available science in these different categories and then, to try to identify some science needs going forward. The timeline of this group, we started with an in‑person meeting in November of last year. We've been having remote meetings since then.

We're hoping to have a webinar series soon this summer, meet again in‑person, and then produce a final report that has a management synthesis, a science synthesis, and a synthesis of those science needs. Again, Aaron Ramirez is the coordinator of this effort and his email address is on the screen if you'd like to get involved or have more questions about this effort in the Pacific Northwest.

Last but not least, I want to mention that there's also this new effort in the Northeast supported by the Northeast Climate Science Center, coordinated by Toni Lyn Morelli, and her email address is on the screen.

The Northeast Refugia Research Coalition has kicked off their group, and they're starting to identify their priorities in terms of species and ecosystems. Those can include Canada lynx, salamanders and vernal pools, moose, Bicknell's thrush, coastal systems, sugar maple.

This is just to say that there are these different regional groups that are working to take some of the existing science around these various different kinds of refugia and really integrate that directly with the needs and priorities of natural resource managers. With that, I am done, and I can take any questions that you guys might have.

John:  Thank you, Jennifer. Hold on, I think I have one right here real quick from Robin O'Malley. He says, "I really appreciate the explicit inclusion of, 'Synthesize available science,' as a precursor activity. While most research includes recognition of the work that has been done before, it often does not include explicit synthesis."

Just a nice note. Thank you, Robin. Toni, I see you have a question. Do you want to...are you on the phone with us?

Toni Lyn:  Great, thanks. Yeah, fantastic presentation, Jen. This was really fun stuff. It's super impressive to see all the work you're doing. My question, or maybe most excitement in the moment, is thinking about your mountain pine beetle work.

You've got this fantastic view of biological response to refugia, and then environmental layers that indicate where you might find refugia in other places, as well as where you use them to explain the biological trends you're seeing. Within what you said, you mentioned thinning. I feel like all of that together is this fantastic opportunity to look at other landscapes.

Even within that national forest, I think it was, in other places to say, OK, well, we can expect that these places will be more resistant to drought and pine beetles, and on top of that, you could do thinning treatments within those to make them even more resistant, because that thinning piece could kind of fall within actions and response, so kind of the next step after you identify refugia.

It was super cool, and I think there's so much opportunity to do some experimental climate adaptation work based on your results here. Thanks for sharing with us.

Jennifer:  Thank you. There's a number of things in what you just said. It's interesting, there seems to be maybe a progression of ideas. Some of the thinking around refugia started in terms of climate modeling and looking at temperature from a very physical process perspective that, of course, is going to be really important.

People started saying, "Well, OK, but what about the biology? What about the interspecies interactions, and what about disturbance, and what about all of these others things that we sort of know play a role in how ecosystems actually respond to environmental change?"

Now, it's exciting to start thinking about, "OK, well, what, what would it look like, refugia from all these different kinds of things?" To try to make it, as much as we can, as holistic as possible, so that we're not just looking at things like temperature, hydrology.

We're starting to bring in all of the ecological processes, which makes it a lot more complex, but hopefully over time, as a scientific community, we can refine it to where it becomes more realistic or more representative of what's going on. That's my hope, anyway.

Toni Lyn:  Absolutely, couldn't agree more. Thanks, Jen.

John:  Absolutely. Now, we have another question from Bill Koon. "It looked like the response of whitebark pine was bimodal, with high probability for refugia in both high‑ and low‑stand densities. Is that correct?"

Jennifer:  Yeah, it's interesting because some of those...

To explain, that was done, this was using presence of refugia as a binary response, and then using boosted regression trees to model these different landscape characteristics. It's interesting, I did see in a number of these, that these bimodal responses, and I had the same question of, "Is that for real?"

That's part of why I really hope that future work can synthesize some of the field work, because I would like to know, is that reflecting a real process on the landscape? Is that true, or from just the remote sensing, is it possible that there's some artifact of how the Landsat was processed? I don't, at this point, want to go too far out there in saying that that is a true phenomenon.

It's hard for me to read, but I think you asked, "How did you measure or estimate stand density?" That's not my data. This is coming from the Forest Service, Forest Health Technology Enterprise team, some of their models of basal area estimates. I don't know if that answers the question.

Some of these results, that's why I wanted to show, "OK, here's the hypothesis, and in some cases, it seemed like, 'Yeah, it's supporting what I thought,' and in other cases, either, 'No', or there's like these bimodal responses."

The next question is like, "OK, is that real? Can we get some field observations to sort of confirm or revise some of these approaches?" I don't know if that answers your question. I hope it does.

John:  OK, thanks. Dan, really quick, asked, "Could you put the Morelli and other references you referred to earlier back on the screen?"

Jennifer:  Yes. Unfortunately, that is like the very first, so I'm going to have to click. I'm going to try to go, really quick, all the way back to the beginning, because these two papers are excellent.

They have a lot of authors on them that have been thinking and publishing about refugia way longer than I have, so if you want more information about this stuff, you can't do better than to start by reading these two papers, in my opinion.

John:  Thank you, Jennifer. Suzanne Baker has a question. "Have you looked at cone‑seed transport based on mechanisms of cone movement and seed release and recruitment?"

Jennifer:  That is an excellent question. The answer is, "No." My work so far has been purely using the remote sensing side of things, using the spectral indices that we can get from remote sensing, but then, in talking with land managers that are more familiar with these systems, of course, that's going to be...

For a refugium to really function in an ecological framework, then these things about transport, dispersal, and then, you get to the question of, "OK, well, what size? You know, what's like an area threshold for a refugium to be viable, and what's the sort of connectivity that you need, in terms of seed transport?"

I think that you raised the next level of questions that I would have. First, from what I've done so far, I would want to say, "OK, we have these remote sensing results. The first thing I would like to see is some field, you know, ground truthing to see if these, they really are where I think they are, and where, you know, the remote sensing says, and whether it's these, the same kinds of processes that support them."

If we develop really high confidence that, "Yeah, these really are what we think they are in terms of refugia." Then, the question you just ask is absolutely the next step.

"OK. Then, what do we know about viability from an ecologic perspective besides the connectivity and, obviously, these interspecies' interactions that are critically important in these habitats?" Given that idea that refugia started, like I'm saying, it can be temperature, physical‑based. It's exciting to see now that we're getting started to ask some of those questions. I think, that's the next frontier.

John:  It doesn't look like we have any more questions coming in through the chat box. I'd like to thank you, Jennifer, for your presentation and once again thank the USGS NCCWSC, that we work with, to put on these presentations.