Mapping Migration with eBird

[Leo Sack] Welcome to today’s webinar on mapping migration with eBird. Thank you all so much for joining us. This is part of the Cornell Lab of Ornithology’s two week long Migration Celebration event. Check out our event web page, at the link that we’re putting into the chat now, for more live events, articles, and resources about migration. My name is Leo Sack and I will be facilitating today’s conversation. Our special guest today is Tom Auer, GIS Developer for the Cornell Lab, who has played a huge role in developing eBird Status and Trends maps. Hey, great to see you, Tom!

[Tom Auer] Great to be here, thanks.

[Leo Sack] Our pleasure, thanks for being here. Before we hear from Tom, I want to go over a few tech announcements. Number one, closed captioning is available. If you would like to see subtitles, please click the Captions button at the bottom of your screen. Number two, we are streaming live to both Zoom and Facebook. Now, if you are watching on the Cornell Lab’s Facebook page, please be aware that we have had some spam attempts in the Facebook comments, so please do not click on any links in the Facebook comments unless they are links posted by the Cornell Lab of Ornithology. Number three, we are going to start with a discussion between Tom and myself, but then we also want to answer questions from the audience. Now if you are watching on Zoom, you can click on the Q&A button located on the bottom of your screen and type your questions into that Q&A window that will pop up. If you see a question already typed into the Q&A that you like, please upvote that question by clicking the thumbs up icon. We’ll be answering some questions verbally, and for others we will be typing in our answers, which you’ll be able to see by clicking over to the “answered” column in that Q&A window. If you are having technical problems with Zoom, or just want to say hi or other comments, please use the Zoom chat window for tech help and other comments, not the Q&A. The Q&A is for questions, chat is for everything else. For those of you watching on Facebook, you can add your questions to the Facebook comments and my colleagues behind the scenes will share those questions with us. So we will try to get to as many of those audience questions as we can. Okay with all that said, let’s get started. Today we want to focus on eBird’s Status and Trends maps, which have been produced as part of using eBird data to model migration. As I mentioned before, Tom here is one of the leading experts behind those maps. So Tom, would you please start us off with a bit of an introduction, both to yourself and to these maps. Briefly, what’s your background, how did you get involved with this project, and what’s your goal in making the Status and Trends maps.

[Tom Auer] Absolutely. Like you said, I’m a GIS Developer with eBird Status and Trends team. I have a background in both biology and geography, but I’ve been a birder for much longer than I’ve been an academic and researcher and practitioner. So I bring that interest to my work as well. I’ve been working now at the Lab for five years on this project, and it’s been really exciting to bring it to the web, bring it to fruition. Because what we really want to do is take eBird data and transform it into a really rich, detailed product that everybody can use, from birders and scientists to conservation practitioners and decision makers. And that’s really the goal of getting these products out into the world.

[Leo Sack] Excellent, thank you. Now let’s back up just a little bit, make sure everyone is on the same page. So we have mentioned that you are doing this work using data from eBird, and eBird is one of the world’s biggest citizen scientist projects. For those who haven’t heard of it before though, Tom could you briefly explain what exactly eBird is. Ultimately where is this data that you are using coming from?

[Tom Auer] Yeah. So eBird data is, like you said, a big citizen scientist project that involves collecting data about birds when people go birding. So we have a mobile app, we have web submission. Where you can go out birding and you start a checklist, you click “start checklist.” And what that checklist is, it’s a list of all the birds you’ve seen while out birding, and the counts associated with those birds. So we need you to help count so we can understand how many birds there are, and how those populations are doing. You go out birding, you submit all the birds you see, the app records — you can enter some information about how long you’ve been out, where you are, how many people you are with, those sorts of effort information that is helpful for researchers and analysts to use the data. Then you submit those checklists, and they go into eBird. There are lots of tools under eBird Explore to check out all the checklists that folks have submitted. But then we take that data and combine it with information about the Earth — landscape-level data, remotely-sensed satellite imagery about the Earth — to then turn it into Status and Trends products. It’s really that whole process of having the community generate such an overwhelming amount of data that makes this possible. Without it we wouldn’t have Status and Trends.

[Leo Sack] So ultimately it is everyday people all over the world who are going out birding and submitting their observations through eBird, that allows you to have the data that you need to do what you are doing, what we’re going to show to people.

[Tom Auer] Yeah, absolutely. Everybody can participate at all levels, all checklists are valuable. No matter how skilled you are, we can use your data, and make it useful for science.

[Leo Sack] Excellent. I think that’s an important point to make. Also, for the sake of time we’re not going to show people the eBird Species Maps, which we considered doing. But I want people to know that they can go to eBird’s Explore section and pull up maps that are basically the raw data, for seeing where people have put in sights of different species. So you can look up Barn Swallow, and see everywhere that anyone has submitted sightings. But that is just, you know, the raw sighting. So what you are doing, Tom — correct me if I’m misunderstanding this — you are combining that with this landscape-level data that you mentioned. So, what the habitats are, and kind of computing it out to, you know, where the birds are likely to live, not strictly just where they are most reported. Because that can be confused with where there are the most people. Is that correct?

[Tom Auer] Yeah, you’ve got it. So eBirders most often live where people are, right. They live in cities and towns, which isn’t always where the birds are. People can get out into, you know, more remote landscapes, but what we need to do with Status and Trends is really fill in the gaps. We need to try to estimate and understand where birds are, and how many of them there are, in places that we don’t have data from. That’s really kind of the fundamental method that’s going on with Status and Trends. We are using statistics and machine learning and computing to really fill in all those gaps, both in space — you know, places that people haven’t visited — but also in time, maybe certain weeks that don’t have as much data. And we can fill those in with this process, so we can create a uniform set of information about each species. And again, using data that everybody submits to eBird.

[Leo Sack] Excellent. Okay. So now we got all that background out of the way, so here is the exciting stuff. Tom, could you please show us eBird Status and Trends maps and tell us more about them?

[Tom Auer] No problem. All right everybody see my screen okay?

[Leo Sack] Yes.

[Tom Auer] All right, so yeah. When you visit eBird.org/science you’ll see a few splashy images that highlight some of our best maps that we have put out here for everyone to see. These are maps that combine all of the seasonal information into one image, to kind of give you something that looks a little bit like a range map, although we have tried to improve upon a range map by adding information about the abundance as well. So not whether a bird could be seen there or not, yes or no, but how many might you expect, if you were to go to a particular place during a particular time period. How many Wood Ducks might you see? We can check out more information here about Wood Ducks. You can see a list of all the products we offer related to each species, here on this Wood Duck page. It starts with abundance animation, which I’ll look at in a moment. Like I said, that combined map here, where red is where the bird is during the breeding season, yellow or green are the migration seasons, blue is the winter. And if a species is present somewhere year-round, that’s in purple. I’ll just focus in on that. You can see that it comes in really rich detail. There is a lot of resolution to this map. And that’s the really key part of our product, is that we’re doing this really fine scale on the ground, so that it can be useful both to birders who may want to know whether they can expect to see a species at that specific location, but also to conservationists, conservation practitioners, researchers, decision makers, to know whether a site might be important for a species or not.

[Leo Sack] And Tom, real quick, so I think you said this, but just to draw it out: The intensity of the color tells you something about the abundance. So for example, with the breeding season area that’s colored red. Places where it is a rich vibrant red is where there are a lot of Wood Ducks in that area. And then places where it fades out into more of a really pale red, are places where Wood Ducks are there but not as common. Correct?

[Tom Auer] Yeah, exactly, that’s totally correct. So if you are looking at this map here, and you are in Minnesota during the height of breeding season, which we defined as being from late April through early August. If you go out at the peak time of day to see Wood Ducks, which is often in the morning. If you go out to a site that has the correct habitat you might be likely to see somewhere between 10 and 30 Wood Ducks, throughout a lot of Minnesota. Whereas if you are out in the western Dakotas or Nebraska, maybe you’ll only see one, if you run into them at the right habitat. So that’s right, the more intense the color here, the more of them you are likely to see. And the really faint colors are places where you may not encounter them all the time.

[Leo Sack] Then you have all this detail. It looks like there’s a gap in the Appalachian Mountains, for example, where you are not going to see Wood Ducks in that area. So you have really taken this fine detail into account, to see things like the spots where there just isn’t the right habitat for them, for example.

[Tom Auer] Exactly. The high elevation parts of the Appalachians are not appropriate habitat for Wood Ducks, and you’re less likely to find them. Of course you wouldn’t expect to see them in forests on top of the great Smoky Mountains. So it’s not a surprising finding, but it’s informative across the entire range of the species.

[Leo Sack] Sure. That’s just a lot more detail than we usually get from your typical field guide range map.

[Tom Auer] Yeah, exactly. Let me take a look here at the abundance animation. This is kind of one of the flagship products of Status and Trends, and one of the things that people are most familiar with. These are animated maps showing where the species is every week of the year. So I’m going to go ahead and click play here. You can see as the time slider moves along, we can see migration of this species progress north in the spring. And then again here now, as fall comes the species is moving south again back into their core wintering zone in the southeastern United States. The colors here are a little bit different, the dark blue and purples indicate where there are the most Wood Ducks likely to be seen on one of those checklists in the morning. And the real pale yellows and oranges are places you are not going to see them all the time in a given week. Maybe only one checklist or one morning out of a week in those places.

[Leo Sack] That’s just entrancing to watch.

[Tom Auer] This tool on our website allows to pause the animation, and take a look. You can also step, by just clicking week by week. So if you are heading out into the field and you are going some place you are not familiar with, and you are maybe keenly interested in Wood Ducks. You could use this to give you a good sense of the types of places, general parts of the region you are heading to, that might be more likely to have Wood Ducks than other places.

[Leo Sack] Excellent, so we could go to this week and see, where are they in their migration just this week?

[Tom Auer] Sure, let’s take a look. Right here about September 20th. We can see there are large flocks of Wood Ducks starting to form in the upper mid-west, so Minnesota, Illinois, Wisconsin, Michigan. As the birds are starting to move south you can see the numbers, the dark colors starting to form in the upper Mississippi river valley, which is a fantastic migratory spot and wintering spot for Wood Ducks. You can start to see them funneling in from the midwest where they breed into their wintering zones down in the southeastern US. One other thing that’s fun to notice about Wood Duck is that the Western populations, those in California and Washington, Oregon, are a little more stationary, they move around a little less. One of the kinds of findings that we can see when we look at these maps and learn some new things about all these different birds.

[Leo Sack] That is fantastic. Now are these maps useful for informing scientific research and conservation efforts? You’ve obviously made the point already that they’re useful for birders, to figure out where and when they want to go out to look for these birds. But how is this helpful for scientific research, for example?

[Tom Auer] Absolutely, so I’m going to jump over to the abundance map for Bald Eagle, just for a second. It is quite colorful. You can see the main regions in lower 48 of the US that the species can be found in throughout the year. So again we have this really fine-detailed information in three kilometers. Let me stop this share. Government agencies and other groups that are, you know, working to manage and protect species, can use this information to help protect species or inform policy. Let me give a great example of that right now. Here we go. So here is a map that is based on that eagle map that I just showed. So using our estimates for abundance estimates for Bald Eagle for throughout the whole year and it is combining them with estimates for Golden Eagle as well. The Fish and Wildlife Service is keen to have information about where eagles may be threatened by wind development. Where is it a risk that eagles might have trouble with wind development? And so what they did is, they used that information that we provided, and classified it into “low exposure to eagles” and “not low exposure to eagles.” So you can see here in green, these are places where there are just not a lot of eagles which are at risk for hitting wind turbines. And the “not low exposure” is places there are enough eagles that it is a real concern. This can then go into the permitting process for wind development. Places where there is “not low exposure” to eagles have to do a lot more to prove that they are not going to take a lot of eagles. Where as maybe in the green areas there is a little faster process for wind development. In the end it is a win-win, because we’re going to help eagles with this.

[Leo Sack] Excellent. So this is actually, so they are taking your Status and Trends data and using it to inform where the eagles are really at risk if there is wind development. So that’s a great example. Okay was there anything else you wanted to tell us about the eagle example here?

[Tom Auer] Not specifically about the eagles, but you know there are other situations where the Status and Trends data can start to help inform different government agencies. This sort of high resolution, this three kilometer information, is kind of unparalleled. They don’t, you know, agencies don’t have as much access to information like this. So we’re just kind of starting to hit the tip of the iceberg in terms of what we’re going to be able to do with this information. Especially as it comes to how bird populations are changing, you know, over the past ten or fifteen years.

[Leo Sack] Absolutely. And speaking of bird populations changing over time, I notice there is a spot on the website that says Trends maps. But they aren’t available yet, it says “update in progress.” So what are Trends maps, and what will those look like when they are ready?

[Tom Auer] Absolutely. So one of the exciting parts of working on Status and Trends is being able to work on research and development of new products, new ways to use eBird data. The new and exciting thing that we’re working on are these Trend maps, and they are in process right now. I’m able to share an example with you today for Wood Thrush. So here is a map of the trends over the past 13 years during the breeding season for Wood Thrush. What you are seeing here is, the size of the dot on the map is portional to how many of this species have been there on average over that period. So large dots mean the species occurs there regularly in number. The tiny small dots are places where the bird is less frequently encountered, has lower abundances. So it gives you a sense of the real core important areas. And then the colors tell you whether the species has been increasing or decreasing. And that’s percent per year over the past 13 years. The intensity of the color gives you an idea of that magnitude. How strongly is it moving in either direction? And so what we see here, is that the species has not been doing as well in the northeast corridor, kind of in the urban zone from Washington DC and Virginia through Pennsylvania, New Jersey, Connecticut, and up towards Boston. But it has been increasing at the western range of its population and throughout the Appalachian mountains. And so results like this are really exciting because we get to see information about trends that are really fine spatial scale for this sort of information. It’s not as detailed as the three kilometer stuff. This is 27 kilometer, so it is a little bit more like county-level. But that’s still really novel for this sort of information. And so there is a real potential to both track birds with this information, but also to start to generate hypotheses about what might be happening with population, so that we can then help them.

[Leo Sack] Gotcha. So in this case, for example, the Wood Thrush loves mature forests. And so you could start looking at, okay, are we losing their preferred breeding habitat in the northeast? Or something like that, and start generating, as you said, those hypotheses about what the issues are and what we could do to help.

[Tom Auer] Yeah, exactly. It seems potentially likely that the continued urban development along that urban corridor is impacting Wood Thrush populations in those locations. You can think of housing developments. And if I look at the Appalachians, I might say those are higher elevation places, perhaps maybe there is a little bit of climate change going on there forcing the birds up slope. But, you know, those are hypotheses that need confirmatory testing, but this gives us some ideas about what to investigate.

[Leo Sack] Excellent. Thank you for showing that. And tell us, when do you think you’ll have these trends maps ready for all of the 600 birds that are in the Status and Trends?

[Tom Auer] I know how many people are excited to see them, so I would love to be able to give a definitive answer, a specific date. But right now, you know, we’re really deeply entrenched in the process of putting together a publication that describes this work. It’s really important to get the peer-reviewed research out into the literature first, before we make this stuff publicly available. So we’re working on that process through this fall and into early winter. So we are hoping that sometime in spring or summer of next year, these maps will all be up on the website.

[Leo Sack] All right. Well thank you for giving us a sneak peek.

[Tom Auer] Absolutely, excited to do it.

[Leo Sack] Let’s move on to some of the questions we have been getting from the audience. One in our Q&A that’s been upvoted a lot. “If I report seeing a bird and ten other people report seeing the same bird, without necessarily knowing it is the same individual bird, how is that useful?” Can you tell us about how eBird accounts for that sort of thing.

[Tom Auer] Yeah. So that’s a really good question. This happens a lot with birding, if a rare bird shows up, everybody runs out and sees it, and submits a checklist because they want it on their list. I do it all the time. But basically, we have so much information, and especially in places like where birds are repeatedly seen, that we can sample that information. And as part of the process of generating these results, we do a lot of randomized sampling. So we’re picking up samples of checklists and creating models out of those, and then reporting that into an average. Basically if ten people go out and see the same bird and we don’t know, that’s okay because we are randomly picking up sample data that gives us variation and variability, that kind of deals with that natural clumping phenomenon that people might have. So keep doing it. Go out and submit those checklists, we have methods for handling those sorts of situations.

[Leo Sack] Excellent, good to know. Tina asks: “The migration maps — the Status maps that you showed us — use data from 2014 to 2018…” Somebody is reading carefully!

[Tom Auer] [LAUGHTER] I guess so.

[Leo Sack] Can you give us a feel for how consistent the timing is, in those migrations from year to year? And will the animations be updated periodically in case climate change is affecting the migration timing?

[Tom Auer] Yep. That’s a fantastic question, and very focused on the detail. So one thing to make clear is that, well for starters, we are committed to producing this information every year. So we are going to be — eBird is collecting 20% more data every year, so we have more data to work with. And yes, things like climate change are acting on birds, and so we want to encompass that in our results. So every year we’re going to produce new results. Towards the end of this year, some time in December I hope, we’re going to put all-new versions of these species up. With even more ranges filled in, better fits, better model results. One of the things to make clear is that, it does say that we use data from 2014 to 2018, but we estimate these results specifically for the year of 2018. So what you are seeing is representative of the year just 2018. It’s not necessarily representative of that whole time period. What’s captured in there is seasonal and yearly variation in timing of migration, or maybe there is water availability that’s driving a species. So yeah, our modeling process is capturing that kind of year-to-year variation, and that’s part of the reason that we’re going to do this every year, so that we get the latest and greatest information about things like migration timing. I know one of the products I’m excited to release in the future is information about phenology, migration timing and arrival. There’s information in here we haven’t tapped into yet, but I’m going to be excited to get it out onto the website in the next few years.

[Leo Sack] Excellent. Sort of a similar theme here — Lots of people have lots of great questions, I’m trying to figure out which way to go. Once that’s been upvoted a lot in our Q&A: “Is there any relationship between eBird data, which gives you the species specific migration data, and radar data used in BirdCast? We had a BirdCast webinar last week. Is there a relationship between the two, or a way to use them together?

[Tom Auer] Yeah. So there is no relationship here in what you are seeing with Status and Trends. There is of course a common reflection between what the radar sees and what birders observe on the ground. There is a correlation there. We have had a collaborator work with us to look at, basically, kind of understanding the relationship between those two. By estimating overall bird biomass in the same way that we model these species, where are the birds based on the eBird data, and then comparing that. This is an active research project. Comparing that to the radar estimates to see if those line up. Are birders and radar good at estimating the same quantity? Because then if that’s true, we can use the eBird data to tell us something about the composition of the radar signal. And the radar signal is much longer than we have eBird data for, it is 25 years long, and it’s also automatic, and covers the lower 48. So there are a lot of really interesting topics that are available at the intersection of those two. So I would suggest checking out some of the latest BirdCast research about that.

[Leo Sack] Excellent. So there is active research going on in that area, but again as you said, it is not reflected necessarily in your Status and Trends maps. The place to look would be the BirdCast website, and their blogs about their research.

[Tom Auer] Absolutely.

[Leo Sack] Okay, excellent. Now back to the Status and Trends maps themselves. You’ve got the color intensity representing relative abundance. How does that relate to the number of birds? You know, how do you determine “relative” versus “absolute” abundance of birds, and is there a method that people should be using to count the number of birds?

[Tom Auer] Right. So, “absolute” abundance in the research field is kind of the holy grail of doing this sort of work. If we could tell you exactly how many birds there were in a three kilometer pixel, we would have a fantastic handle on how to help birds, even better than we can with this information. However, basically what happens is that birds’ ability to be detected changes throughout the course of the year. You can imagine those confusing fall warblers versus when they’re singing on territory in the spring, it’s much harder to detect them and properly identify them and find them all. Because of that change throughout the year, we don’t know exactly how many birds are at any given location at one point in the year. And so it’s a continuing ongoing research struggle to try to estimate that absolute abundance. Man would we love to do that. But, for now this abundance index, this abundance measure is the best thing we’ve got, and it is powerful.

[Leo Sack] Okay. Now we’re starting to run short on time. I see a lot of questions that are kind of all in a similar theme here, but it is not something that I think is likely to show up right away in your Status and Trends maps. So I’m going to kind of throw all of these out here in one big bundled question, and see if you’ve got something you can speak to on this theme, Tom. So there are lots of questions along the lines of: “How do you think southbound migration this fall will be affected by all of the hurricanes this season?” There’s also questions about the wildfires out west. Irruptions, somebody mentioned irruptions of birds, like lots of one species appearing or moving to an unusual area due to food sources. So you’ve got all of these different things that could happen, you know, this season being different from other years for various reasons. So, how well does eBird track that? And to what extent does that show up in the Status and Trends maps?

[Tom Auer] Yeah, so you make a good point. The hurricane situation right now is not going to be reflected on what you see on the website right now. But, I will say that when there are major shifts — you can think of the finch irruptions that happen in the winter. We are starting to capture that well in these Status and Trends products. If you look, for example, at the Evening Grosbeak that’s currently on the website, it really accurately reflects the experience that people had during that one given winter. So you know, we can really understand what’s happening on a year-by-year, and if something is of a big enough magnitude that birders are detecting it, it will reflect in Status and Trends products on a year-to-year basis. You know, Dickcissel irruptions into the mid-west. Or if there are, you know, some sort of major hurricane issues, sometimes we have seen birds kind of get pushed further north as a result of a hurricane. So we are capturing that, and we are always trying to do a better job of grabbing those sorts of seasonal dynamics and things like that in there. So we want information about fire history, or rainfall or climate, we’re investigating how to use these in our products. That will give us an even better handle on those sorts of issues.

[Leo Sack] Excellent. So the Status and Trends maps you showed us today are more average data that won’t necessarily show up those things immediately, but eBird is collecting all that data and it is all scientifically useful.

[Tom Auer] Absolutely.

[Leo Sack] Okay. So this is all so interesting, we could do this all day, but we are at the end of our scheduled time. I want to be respectful of everyone’s time, so we’re going to wrap it up here. Tom, thank you so much for talking with us today, and thanks as well for all of your hard work on these amazing Status and Trends maps. I’m really looking forward to the Trends showing up there.

[Tom Auer] Absolutely, this was a blast. Thanks for having me.

[Leo Sack] I also want to thank our audience for joining us today. We had a great turnout, that’s really wonderful to see. If we didn’t get to your question today please email us, and we will be happy to follow-up with you more directly. So I’m going to share my screen here. For general questions about the Cornell Lab of Ornithology or about our public programs, bird ID help — pretty much any random question about birds — please email our public information team at CornellBirds@Cornell.edu. For more technical questions about eBird you can email eBird@cornell.edu. And finally, don’t forget to explore the Status and Trends maps we discussed today, which are all at eBird.org/science. So that’s our show. I hope you all enjoyed it, and I hope you’ll make use of these awesome maps to help you understand migration and inform your own birding adventures, as well as knowing that they are informing our science and conservation efforts. And keep submitting your data to eBird because it is your observations that make this research possible. So happy birding everyone. Thanks for joining us.

End of transcript