SHACKLETON MEDAL 2026 // SHORTLIST NOMINEE: IN CONVERSATION WITH ELLEN BOWLER

“We realised that the government of Nunavut had amazing tracking data for the caribou. It’s been collected over decades, so we had all of these examples of caribou crossing the sea ice that we could combine with the sea ice satellite imagery.”

Your report – which combines satellite observations, GPS tracking, AI forecasting, and local expertise to predict caribou migrations across sea ice – has been heralded as a game-changer. People are particularly excited about the way you combined AI with Indigenous knowledge on the ground to protect migration routes against factors like icebreakers.

It was a really amazing project to take part in. I joined the Cambridge AI Lab, where IceNet [a deep learning system that forecasts sea-ice] was a tool that had already been developed and there was a lot of interest in how the forecast could be used to inform conservation. The World Wildlife Fund was instrumental in driving that forward and finding partners with whom we could do a case study. So they paired us up with researchers at the government of Nunavut, who were experts on caribou migration. It quickly seemed obvious that having early warning of the timing of the migration would be beneficial for mitigating negative impacts for caribou.

How did you coordinate the research?

We did lots of sessions over Zoom with the government of Nunavut’s researchers. Then they went and talked in person to the community and people who were stakeholders. We’re hoping that the next step will be for us to go there ourselves.

You started off focusing on albatrosses.

For my PhD, I was based in [Cambridge’s] computer science department. My background is maths and computing and I was really interested in applying it to more conservation and wildlife. I was lucky enough to do a project with Peter [Fretwell, the world’s leading emperor penguin expert] on wandering albatross detection in satellite imagery. My focus was on the population on South Georgia. The aim was to get data from areas where you usually couldn't, and then automate the counts with machine learning. I continue to be involved with a few of those projects. I’ve been working with another British Antarctic Survey researcher on elephant seal surveys on South Georgia. Beyond that, there’s a big project there monitoring the impacts of bird flu.

Your specialism was originally maths. How did you transition to this kind of work?

I realised the pure math side of things wasn't for me. I was much more interested in applied things. I did a Master’s in computational methods in ecology and evolution, and that's what got me very interested in satellite imagery. I found it really interesting that you could get information out of a picture. While there’s often a lot of information in imagery, it’s not always easy to get tools to pull the data from the images. That’s why I wanted to work with Peter on the satellite imaging project, so I could work with the right sensing data.

You’ve talked about the fact that IceNet was already developed. So your achievement was to extend the way in which it could be used. How did you go about this?

IceNet forecasts ice across the whole Arctic, which is obviously a broad scale. We wanted to zoom in on a specific region to create a case study and look at ways in which the model could be improved. So we were looking, for instance, at whether it was high enough resolution. Was it doing well in forecasting ice cover for a particular area, or was it overcompensating? Questions like that. We realised that the government of Nunavut had amazing tracking data for the caribou. It’s really unusual, actually, to have such good data on the migrations. It’s been collected over decades, so we had all of these examples of caribou crossing the sea ice, and then we had the sea ice satellite imagery. A lot of my work was linking those two data sets together and trying to establish at what point in the sea ice formation the caribou would decide it was safe enough to go.

Are they adjusting to shifts caused by climate change or are a lot of them dying because the sea ice isn't firm enough?

There are some sad stories about caribou drowning because the ice is not thick enough. They’re also now getting stuck on Victoria Island for longer than they would have done. The migrations have definitely got later, which can mean they might run out of food, or they might be more vulnerable to predation. So yes, there’s a huge amount of pressure on the herd, not just from the risks of the sea ice crossing, but from disease and other factors.

What are those other factors? 

There aren’t just shifts in the sea ice but also potentially in their foraging habitat. That obviously has an effect on the distances they might need to travel to get their food. Caribou are also increasingly encountering rain-on-snow events, which makes the surface of the snow a completely solid crust. That means they can’t get through it to eat, which puts them in danger of starving.

How would you like this project to develop? 

It would be good to work much more closely with the communities themselves. They’re really the ones who would be making decisions and communicating with icebreaking vessels. They’ve already got very innovative conservation measures in that region that they're working on with the government. One of the concerns over icebreaking vessels in the region is that people can be out on the ice when the vessels are in the area. That means that when the icebreakers go through, these people could get cut off from the land. So alongside the problems involved with disrupting caribou routes, there’s a big human safety issue.