What did I find?

Normally, methods are supposed to precede results. You may ask—as per Pink Floyd’s The Wall—“How can you have any pudding if you don’t eat yer meat?” But since I’m a grown man and not a British schoolboy in a 1979 music video, I can do what I want. If you’re interested in how I got these results, I’m sharing that later in the post.

Moose time!

So did the intense moose foraging affect Isle Royale’s bird communities? Well . . . uh, we’re still not quite sure about that. When I ran scenarios with and without the island-wide moose population size, they had almost the exact same predictive power¹. Their quality was so similar, that we can’t really say which is more accurate. However, to me this is still surprising: the idea that the number of moose across a 200 square mile island can say anything about which birds you find in one given spot is pretty crazy. Honestly, this is better than I expected.

If we look at the scenarios incorporating moose (which are equally as valid as the moose-less versions), we find a few cool things. Firstly, it looks like the average nesting height of birds increased from ~6 feet when moose were rare to ~7.5 feet when moose were abundant. Not a huge change, but perhaps moose are disrupting ground-nesting birds! I also found that many birds are less likely to be found when the current moose population is high, but are more common when the moose population seven years prior was high. (Interestingly, birds that nest in deciduous shrubs have the opposite pattern, perhaps because moose grazing makes the environment more shrubby?)

But there’s more!

Since we kinda expected that moose population would be incredibly significant, I had a backup plan. Isle Royale has been collecting bird data for decades, but all the parks in the Great Lakes region have been counting birds since around 2010. That’s when the Great Lakes Inventory and Monitoring Network (GLKN) stepped in and made sure everyone was on the same page, especially with how parks should collect and record similar types of bird data. With this data at my fingertips, I decided to compare how bird communities were changing between parks and over time.

Below, I have included a few example tidbits of what I found, graphed for your viewing pleasure!

But perhaps the most important takeaway: I found that the parks are not immune to the recent continent-wide bird population declines. If a species is declining across North America, it is likely declining even in these near-pristine parks. It’s sad, but it makes sense: birds that migrate here from as far away as southern Brazil will spend time in many countries. Protecting their breeding grounds isn’t enough to ensure their population stays strong.

What did I do?

Ah yes, time for “the boring part.”  I’ll keep it simple though! For the analysis, I learned how to use Hierarchical Modelling of Species Communities (HMSC), which is, to quote its creator, “a model-based approach for analyzing community ecological data.” Essentially HMSC is a way of investigating which species you find where and why. The idea is actually pretty simple: species are usually found in environments that match their traits. For example, if one area has more shrubby trees, you might expect to find more shrub-nesting birds there, right?

To use this approach, I needed bird counts, bird traits, and environmental measurements. The bird surveys fulfilled the first requirement, but let me tell you, 25 years worth of field data is NEVER in good shape. Typos, misidentifications, and duplicates abounded, and I had to fix this data up before it would work in the analysis. For the bird traits, I talked with my ornithologist collaborators and used data found in the recent, incredibly famous “Hey all the birds are dying, you guys!” paper². Lastly, took the park’s vegetation survey and island-wide moose counts to describe the environment.

There is much, much more to these models, but my brevity is for your benefit, dear reader. The full story is one full of sound and fury, but is ultimately, signifying nothing. Ahem, anyway. Every time I wanted to create a scenario, I had to run HMSC on my computer for hours and hours. In fact, these models started taking so long that I eventually had to learn how to run them on my university’s super-computers. (Made me feel like an elite hacker or something cool like that)

Anyway, it was a grueling, yet rewarding summer experience! If you have any questions or just want to get in touch, shoot me an email or hit me up on Twitter.

1. The WAIC scores were virtually identical between models that included moose population as a parameter and those that didn’t. As a rule of thumb, we say that if models are within two AIC points of each other, they both have “substantial support.” In this case, ΔAIC was 0.008…
2. Fun fact: while trying to use their data for my own analysis, I found a small error in the paper! The authors had forgot to include a column in their supplementary table that I needed. A quick chat with my man Kenny (listen, we go waaaay back) fixed the issue and got me the required data.

We were gliding over the glassy water when the first rosy-fingers of dawn began to color the sky. I was hunched down against the wind whipping across our bow, serving as human ballast. Jared and Scott sat further back and were more sheltered from the cold, but I didn’t mind. Using my weight to balance the boat would clearly be my most significant contribution to the team.

After all, I was finally on Isle Royale, and I was ostensibly there to help the bird surveying team. Jared was our ornithology expert, but he was new to both the island and the survey routes. Scott had spent the last five summers on these trails and—although he initially tried to deny it—was also clearly skilled at identifying bird song. That left me the roles of team photographer, continually impressed audience, and … boat-balancer.

The Mission

After our pre-dawn disembarkment, (boats take on the role of automobiles on Isle Royale) we hit the trail. We would hike briskly for about 250 meters before pausing. Over the next five minutes, Jared would enter some Wuxin-like mental state and slowly piece apart the location and species of each individual bird responsible for the chorus of birdsong surrounding us. Scott would record Jared’s dictations and also voice his own observations. I would stand in awe of their skills, desperately try to hear what they were pointing out, and occasionally crouch down to look at some cool bugs I spotted.

By hiking the same trails year-after-year and recording the birdsong at the same points, we can get an idea of what’s happening to the bird community of Isle Royale. Also, these transects were designed to go through some very different ecosystems. On the island, there are dense hardwood forests, wet marshes, rocky meadows, and conifer groves all sitting side-by-side. In fact, we passed through all of these on our relatively short, four-hour hike. Hopefully, I will be able to leverage the data we collected here to understand how these different habitats inform bird nesting sites.

Ecosystem engineers

But these habitats were clearly not permanent, immutable aspects of the island. Along our survey, it became obvious that there were two culprits rearranging the landscape: moose and beavers. Whole clearings of young trees were sliced across at a precise five foot height, as if shorn by some giant, heavenly weed-whacker. Moose can be very thorough diners, it would seem.

Even more obvious were the beavers. Each dam would massively change the landscape. In fact, some of the trails we needed to cross had recently disappeared under rising water. In less than a year, two adult beavers had turned a trickling, half-pace-wide stream into something that resembled a lake. Pointy, rodent-chewed stumps dotted every wetland we walked through.

How do the birds react to such “ecosystem engineering”? I’ve decided, if I have enough time, to try and include the beaver population in my statistical model alongside moose. Both of these mammals seem to be altering habitats, and it will be important to understand how birds react to the change. Plus, as the wolf population increases at the expensive of these herbivores, we will be able to make predictions of future bird communities!

Bonus wolf print photo:

Ahhh, to the seasoned ornithologist—of which I definitely am one—the morning chorus of birdsong is as familiar as an old friend. There, can you hear that cooing? That must be a dove.

And what about the high-pitched, chirpy noise?  Because of my very experienced ear, I can clearly tell, um, that… it is… probably a songbird? Yes, a… red-headed songbird. Very common in these parts, I must say.

Oh, and you’re curious about that twittery, raspy, whoopy thing too? Uhhh… a fluffy-backed–no, a horned… um, I think it’s called a.. rough-faced… bush-babbler.

About me

Okay, listen: I, Andrew Burchill, behavioral ecologist that I am, may or may not be an ornithology expert. In fact, I may or may not know anything about birds whatsoever. Yet I am here in Houghton, Michigan to suss out the secret mechanisms that build bird communities on Isle Royale National Park.

Normally, I live as a PhD Candidate in Arizona studying the collective foraging behavior of ant colonies. (Essentially, I ask how huge families of ants manage to coordinate healthy meal-planning and grocery shopping.) I spend most of my time staring at these tiny black ants, wondering what they’re doing:

Although studying ants and birds is clearly different—you know, with the feathers and the beaks and what have you—all science is rooted in data. Years of teaching myself statistics and analyzing gigantic datasets have made me a whiz at examining data.

About Isle Royale

Isle Royale National Park is—you guessed it—an island, or technically a series of islands in Lake Superior. Ignoring its rich and varied history, I can say that Isle Royale is basically a big, mostly-untouched island full of birds and moose and beavers and (kinda) wolves. Every ecologist will hear about this special place during their education. The island has functioned almost like a giant, contained lab experiment for the last 60 years, where scientists monitored how predators (the wolves) interact with prey (moose). But because of inbreeding, the wolf population essentially dwindled away, and efforts are ongoing to transplant more wolves to the island.

Although technically considered a part of Michigan, Isle Royale is much closer to Canada than anywhere else. In fact, the NPS headquarters in Houghton is a six-hour ferry ride away from the island! It certainly isn’t the easiest place to get to. As the Isle Royale staff like to say, it is the least-visited but most-returned-to national park in the lower 48 states.

About the project

And speaking of “least-visited,” I personally have not yet actually been on the island yet. Because of COVID, both the official housing on the island and the official ferry to the island is operating at half-capacity. But that’s okay because there is still PLENTY of data that needs to be analyzed!

Dedicated volunteers have been surveying breeding birds every spring since the mid-1990s. They walk specific trails, pausing at pre-defined spots and listening for bird calls. Unlike me, they are so experienced that they can identify the species based on the songs and calls around them. It’s my job to take this mountain of data and uncover how the bird communities on Isle Royale have been changing over time. This is especially important, because bird populations across North America are plummeting. My work will let us know if the songbirds in this amazing national park are also in danger!