.png){kind=icon}
Name: Peter Fretwell
Organisation: British Antarctic Survey
Role: Geographical Information Officer
Nationality and main countries worked: British; works in Antarctica
Why is establishing species’ distribution and movement important in applied ecology?
If we are to understand species’ population sizes, as well as how and why that population might change over time, establishing the baseline distribution and any annual movements is critical. It is impossible to get any handle on population dynamics without a good knowledge of where individuals are and where they move during their lives.
What is your day-to-day job?
I’ve been lucky to be at the forefront of what is becoming a revolution in applied ecology: using remotely-sensed images, such as satellite images, to study the distribution and movement of individual species. The use of satellite remote sensing of wildlife has developed rapidly over the past five years, mainly because of the ease of access to imagery and the increasing resolution of satellite censors. I have been trying to apply this new technology to a range of species to help us estimate abundance, distribution and movement for ecology and conservation purposes, mainly in the Antarctic and Southern Ocean.
What is your most interesting recent project and why?
Show most people a picture of an emperor penguin and the chances are good that they will recognise it. The species is charismatic, unique, and well-known in popular culture. However, that familiarity belies the fact that we actually know very little about the species’ ecology. It breeds in some of the most inhospitable places on earth – inside the Antarctic Circle – at one of the most extreme times of the year when the Antarctic winter means that temperatures drops to -50˚c and the continent remains in total darkness.
Emperor penguins breed on sea-ice rather than on continental Antarctica. Sea-ice is a habitat that is dynamic and constantly changing: at its maximum annual extent it doubles the size of the continent but then shrinks back to almost nothing in the Antarctic summer. Breeding in this environment means that emperors are very vulnerable to climate change impacts on sea-ice. Their entire breeding cycle – and thus the population dynamics of the whole species – depends on the stability of the habitat; a stability that modelling shows is under considerable threat. Crucially, though, we have very limited knowledge of their distribution because the very fact they breed on sea-ice makes them very hard to find. Between 1948 and 1958, the Falkland Island Dependency Survey (the organisation that became the British Antarctic Survey) monitored one colony of penguins in the Dion Islands annually. That ended in tragedy in 1958 when a storm broke up the sea-ice with the loss of the research team. For the next 50 years, British Antarctic Survey, one of the leading research organisations dedicated to one geographical area, did no research on the continent’s most charismatic animal.
In 2008, I was working for the British Antarctic Survey in a cartography (mapping) role. One of the things we did was to create route maps for the aircraft to avoided dangerous areas and no-fly zones, which included penguin colonies. For the colonies of smaller penguins, such as Adélies and chinstrap, that was no problem. However, the data were very poor for emperor penguins. For example, we knew that there were two emperor penguin colonies on the Brunt Ice Shelf, but we had no idea where on the ice shelf they were. As each colony might only occupy a few hundred square metres, and the iceshelf is the size of Lincolnshire, this was a problem. When looking at the Landsat satellite images, I saw a large brown stain on the sea-ice. Now, sea-ice is literally that: frozen sea with no impurities. I started to put ideas together and wondered whether this brown strain was evidence of one of the missing penguin colonies. I looked at the other possible penguin site, and, sure enough, there was another brown stain. At that time, there were 25 known colonies and a few others that were considered probable. Helped by the fact that Landsat had just made images freely available, we remotely surveyed around 90% of the coastline. At the end of our scrutiny in 2009, we had found 38 colonies through these brown stains, which turned out to be guano (bird poo). We were literately sensing penguins from space.
The issue that we still had, though, was the resolution of the images. Each pixel covered an area 30m by 30m so the images were rather coarse. We then tasked a very high resolution (VHR) satellite, Quickbird 2, to fly over the potential colonies later in 2009. The pixels on the images covered by this satellite cover an area of just 60cm by 60cm, but it has to be specifically sent to a target area. On the VHR images, we could actually count individual penguins when they were travelling between the colony and the sea. When they were in a huddle this was much harder, but we developed a computer technique to look at the density of the penguin huddle by digitally analysing the “penguin pixels” relative to the background of snow, ice and guano, and we estimated numbers from that. We analysed every emperor colony in a single breeding season to get a robust estimate. The world’s population estimate for emperor penguins went from 310,000 to 595,000 in one year by virtue of the world’s first satellite census of any species, anywhere.
What’s been best part of this particular project?
Well, it’s quite nice to be personally responsible for finding half of the world’s emperor penguins! More seriously, though, some quite unexpected things have come out of this project. For example, we had another missing penguin colony. On his exploration on Antarctica in 1893, Carl Anton Larson documented numerous emperor penguins near the Jason Peninsula. We couldn’t find the colony until we looked not just at the sea-ice but at the ice shelf beyond it… …and there were our missing penguins. Nobody knew penguins could even get onto continental Antarctica, so we discovered new breeding behaviour using a satellite. Pretty amazing.
What do you see as the main challenges in your field and how can they be overcome?
Ground truthing is an issue; traditional ground base or aerial surveys are done over a period of time, but with satellites you get a snapshot and we need to understand the implications of this when assessing populations. We also need to understand the limitations of the imagery, there are some terrains that satellites just aren’t suitable for; forests, rough seas and cliffs spring to mind, but that leaves a lot of other areas that might be suitable. The cost of the satellite imagery is definitely a concern at the moment but hopefully, like most technology this will come down over time. The datasets can be very large so there are problems of data storage and processing. Often region assessments require thousands of square kilometres of imagery to be assessed. Looking for signs of animals that might only be 30cm across is something that takes too long manually so most of the work is aiming towards automated analysis, developing machine learning intelligent search protocols that will automatically count wildlife.
Crystal ball time: mapping technology and data possibilities have changed a lot over the last 5-10 years. What technological advances do you think the next 5-10 years will bring?
Even higher resolution imagery meaning we can look at smaller organisms, thermal imagery, crowd sourcing and cheaper images for conservation purposes are all on the horizon. A new even higher resolution satellite, giving us imagery at 30cm resolution has gone up recently, so now we should be able to see even more species. I hope that we can extend the technology out to monitoring many more types of animal, not just in the polar regions but in any non-forested areas where there is a conservation need around the world.
What next for you, and why?
We need to assess the penguin population trends over initially a 10 year period and ideally more in the future. We have a snapshot but we now need to continue this work to see how this changes over time. More generally, we are also expanding this technique to assess the distribution and population of other species. For example, we started looking at whales – specifically the southern right whale. My theory on this was simple: we can see penguins by satellite and they are quite small; whales are bigger so we should be able to see them too. But of course it’s not quite that simple! We looked at an area of Argentina which has a very high density of whales and managed to set up an automated procedure in GIS software that could count whales automatically by looking at the difference between the whale “pixel signatures” relative to the background (sea) signature.
Finally, how did you get into mapping and spatial analysis and what advice would to others?
Luck mainly, although you have to have the right skill set and the attitude to take the opportunities when they turn up. I am a very much an optimist, and pretty tenacious when I think that I have a good idea, which I think has helped. That said, I’ve found that you have to have a thick skin if you want to progress in science, battling for funding and submitting papers can be a disappointing business.
As for how I got into mapping and spatial analysis, in my case I did a degree in Geography then went on to study Quaternary Geomorphology. That mainly involved mapping and spatial analysis, and from there I took a very low paid temporary job at BAS, which basically was just looking after their map collection. I did a few maps for them and they liked what I produced and I managed to get a job as a junior cartographer. I started applying my science background to a few projects and became an expert at GIS and spatial analysis mainly through on-the-job experience. Then the penguins turned up and since then I have been concentrating on remote sensing of wildlife, publishing scientific research and exploring new applications. I am now a senior scientist at British Antarctic Survey with around fifty published papers and over a thousand citations.