Q & A: Big data meets big (polar) bears

By Riley Tjosvold

Data about our environment has become more detailed and precise, and it continues to be combined in unexpected ways as we are learning more about animal behaviour than could ever be achieved through observation alone.

Ron Togunov, a PhD student at UBC’s Institute for the Oceans and Fisheries, understands this well. He’s been working with tracking data collected from GPS collars on polar bears, combined with data about weather patterns and wind direction in western Hudson Bay, Canada.

His goal: to learn how polar bears find food and understand how they use wind to guide them in their search for prey.

To wrestle with complex bear movement datasets, Togunov had to devise new statistical methods, some of which he believes can be applied to other studies on animal movement.

We talked with Togunov about the insights he’s been getting into polar bear foraging behaviour.

How do polar bears use wind to find prey?

The polar bear research started by testing the hypothesis that polar bears move crosswind to locate prey. If the bears use their sense of smell to search for something, what’s the optimal way to move through the environment to locate it? Assuming that you smell nothing at all, you have no signal, no idea where prey are, how do you move through the landscape to try and find something? The only way to find something is if it happened to be directly upwind of you. By moving crosswind you’re constantly passing through new streams of air and learning about everything that’s upwind of your path.

What have you learned about polar bear behaviour since starting this project?

The biggest insight is that there is a lot of this crosswind movement that hadn’t been shown previously in a large carnivore.

More recently, using the higher resolution data we have, it seems that this behaviour is actually more prevalent than I suspected: Most of the time that the bear is actively moving, it is moving crosswind — to the point that it’s almost making the analysis difficult because I was hoping to take that active movement and break it down into different behaviours.

What statistical strategies have you developed to get the information you need out of this complex bear movement data?

Not all of the behaviours we see in the entire movement track of a polar bear have the same ecological significance. Some might be more important than others, for example foraging vs. travelling vs. resting. If I want to analyze the behaviours individually, I have to be able to classify them. The method that I propose builds on a popular model for classifying and segmenting a track into different behaviours.

Specifically, the method helps identify the behaviours with directional “biases.” Biases here refer to when an animal navigates at some angle relative to something else — for example, a polar bear moving crosswind or then change its direction towards a source of prey. You might not know the angle of a bias; cross-wind is not necessarily 90 degrees — it might have a slight upwind or downwind bias that we don’t know about. Because of the effect of wind on sea ice drift, when a bear is stationary (for example when it’s resting or still hunting), its track also moves at some angle to the right of wind (10° to 25° depending on latitude).

It’s a broadly applicable method, and I’d love for other researchers to consider whether their own study species may have a behaviour with directional bias that hasn’t been investigated.

Why is it important that we understand how polar bears move through their environment?

The main question we’re trying to answer is: What is the most important habitat for the animal? What should we be conserving?

Typically we use telemetry data to see where the animal goes, and we say all of that area is important. But animals are engaged in different behaviours, and different behaviours have different value to those animals. I propose that it is more important to identify the most critical pieces within that movement data — and for polar bears, I think that’s where they’re foraging. The amount of energy they have is the foundation that they need to sustain themselves, reproduce, and survive.

The study “Characterising menotactic behaviours in movement data using hidden Markov models” was published in Methods in Ecology and Evolution. “Windscapes and olfactory foraging in a large carnivore” was published in Scientific Reports.

Tags: Animal movement, IOF students, polar bears, Research, SERG, statistical ecology