A remote rainforest near Knight Inlet on the British Columbia coast has been kitted out with dozens of discreetly placed cameras. When a lone figure wanders into view, one of the devices starts recording as the subject lumbers along a well-used trail, causally checking out her surroundings. The footage is sent to a computer for analysis by artificial intelligence. The algorithm locks onto her eyes and nose, comparing her face to a database of individuals who frequent the area. In moments, it identifies her as Known Female F016.
The system’s operators are already familiar with F016. They call her Flora. They know where she lives, what she likes to eat, that she’s a mother and that her family is a big deal in these parts.
They know this because Flora is a grizzly bear, and the cameras are being used by a research initiative called the BearID Project. The researchers are testing non-invasive techniques to track dozens of grizzlies that hang around an estuary about 250 kilometres north of Vancouver to unpick the behaviour of this iconic yet enigmatic species.
“You get to understand that this isn’t just an animal with a number, it’s an animal with a whole life history behind it,” says Melanie Clapham, a conservation biologist at the University of Victoria who is involved in the project. But IDing bears is tricky. They change appearance as they gain and lose weight throughout the year, and even an experienced researcher might struggle to pick out a particular individual from a lineup of grizzlies.
After several years of development, the AI system now gets it right 84 percent of the time and Clapham says that it should improve further as it continues to gather data. The researchers hope better tracking will deepen their understanding of questions like how far the animals roam, where they go and how they interact with one another. That might provide information about particularly important habitat areas. The technique might also be used to identify bears that are repeatedly drawn to towns and cities.
And, ultimately, such information could provide clues about a far larger concern: What’s in store for these magnificent creatures as human development and climate change encroach on their world?
Facial-recognition software might seem an unusual approach to studying grizzlies, but the BearID Project reflects a rapid embrace of new technologies by conservationists.
In the past few years, drones equipped with cameras or LiDAR scanners have become favoured tools for biologists to discreetly study animal populations or conduct plant surveys. Canadian companies, such as Flash Forest are using them to map areas and replant trees after wildfires. And they’re being used on the seas as well as in the air. Last year, environmental organization WildAid and B.C.–based OpenOcean Robotics tested a small unmanned boat to monitor illegal fishing in a humpback whale reserve near Hawaii. It stayed on patrol for 25 days.
Megan Leslie, CEO of World Wildlife Fund Canada, says that the availability and affordability of some new technologies have put them within reach of non-profits.
“Wildlife conservation has always looked to use technology in interesting and innovative ways, but it just hasn’t been easy enough in the past,” she says. “Now, anyone can go to a store and pick up a drone. The technology out there is much more sophisticated and user friendly.”
Technologies like drones, artificial intelligence and satellite data are enabling environmental organizations to gather the kinds of data that would previously have taken enormous resources to collate.
With its Walrus from Space project, the WWF has created a global network of more than 11,000 virtual wildlife spotters, who download satellite images of polar regions and mark any walruses they see. The aim is to estimate the sizes of walrus populations and identify spots where they gather and need to be protected from disturbance by shipping.
The charity is also experimenting with using more proactive technologies to guard sensitive areas. It is currently studying whether artificial intelligence can pick out whale sounds detected by hydrophone sensors in the waters off B.C. and potentially warn nearby ships in real time to slow down and reduce their engine noise.
But if conservationists’ abilities are growing, the challenges they face are, too.
At the UN biodiversity summit in Montreal in December, more than 100 countries signed up to a pledge to protect 30 percent of land and water by 2030. Canada’s environment minister, Steven Guilbeault, likened the agreement to the landmark Paris accord that has spurred action on climate change.
But the international deal also reflects the vast scale of the threats to natural habitats. Animal and plant species have vanished at such an alarming rate over the past century that some scientists believe we are now living through Earth’s sixth known mass extinction. And Canada is no exception.
The most recent governmental Wild Species report on the nation’s biodiversity found that one in five studied species is at some degree of risk. Last winter, Canada saw the largest loss of commercial beehives in the past 20 years, with 51 percent of those in Alberta being wiped out. Marc-André Roberge, co-founder of Nectar, an app that helps beekeepers manage their hives, says wild bee populations — which play vital roles as pollinators for plants — likely fared as badly or even worse.
“If you’ve got crumbling honeybee populations, you’ve also got crumbling populations of Mason bees and bumblebees and wild native bees,” he says. “They are facing similar threats: depleted land, loss of habitats, toxic environments based on the agents that are being used in agriculture and climate change that is depleting the environment itself and creating instability in the forage available to the bees.”
Efforts to reverse the loss of biodiversity are hampered by a lack of understanding about what we are trying to save. The Wild Species report is the most comprehensive to date, but its authors lacked sufficient information on two-thirds of the 80,000 or so species believed to be in Canada to even determine if they are threatened. In many cases, we don’t have a detailed picture of where habitats are, especially if they’re smaller areas like ponds or groups of trees that may form links in a chain but are easy to miss on their own.
Brandon Palin, a senior director at Ecopia, a digital mapping company, points out that the natural world is much harder to map than the human one at fine scales. “If you’re drawing the boundary of a simple building, you could essentially have one point at each of the four corners,” he says. “That’s a simple polygon compared to a tree, where you might have thousands of points around it.”
Because of the intricacy this presents, habitats have often been charted in fairly broad brush strokes from satellite imagery.
Ecopia usually creates maps of urban areas for planners or insurers using artificial intelligence to extract detailed information from high-resolution aerial images. But it has also turned the technology to the natural environment and has mapped marsh areas of the prairies that are important for birds. This summer it landed a $10-million contract with the National Oceanic and Atmospheric Administration to help the U.S. prepare for climate change and it recently mapped ponds and other bodies of water around Tampa, Fla., in minute detail.
“You have ponds established in these communities that are extremely important for biodiversity and bird habitats,” says Palin. “But that level of detail hasn’t existed in the U.S. National Wetland Inventory before.” More detailed maps should help planners to protect sensitive habitats.
All this technology is generating vast amounts of information for conservationists to sift through. In some cases that presents its own issues — the BearID software has logged so many images of grizzlies that Clapham has to ship hard drives across the country to share them with her collaborators. But it also captures moments that highlight the natural wonder of our planet.
A few years ago, a WWF drone monitoring a population of narwhals recorded footage that may finally explain why they have their enormously long tusks. For years ecologists have speculated that the tusk is used for fighting, mating or spearing fish, but as the drone passed overhead it caught one of the animals using it in a previously unimagined way. The creature swam up to a school of cod and struck one with its tusk. The cod was momentarily disoriented. The immobilized fish became an easy meal.
“Because the drone was right above them, we could see exactly what they were doing,” says Leslie. “We got this incredible footage. Nobody has ever known that before.”
Want to learn more about what it will take to build a sustainable and prosperous planet? Meet the new cohort of Mission from MaRS Climate Champions.
Photography: Melanie Clapham (bear image); BearID Project (bear video); Yulia Bogomolova/WWF-Russia (walrus image)