The great white data hunt
- Creado en Jueves, 09 Marzo 2017 17:41
Researchers at Stellenbosch University have created an ingenious way to track great white sharks.
For most people, working with sharks would be enough to make them jumpy, or at least a little bit anxious.
These emotions are not what marine biologist Sara Andreotti feels. A researcher at Stellenbosch University in South Africa, Andreotti is tasked with quite the undertaking: Collecting data about great white sharks.
“During the sample collection and lab work my thoughts are very much focused on the task at hand and the results,” Andreotti says. “Sometimes I’m happy when I get it right and sometimes I’m thinking of swear words (in multiple languages) when something goes wrong.”
To track white shark sightings off the coast of South Africa, Andreotti collaborated with an applied mathematician and a software developer from Stellenbosch University to create Identifin, a searchable database.
Their research was recently published in Marine Biodiversity.
“The need to identify individual white sharks made me discuss a photographic identification technique with Michael Rutzen, a world-renowned shark conservationist,” Andreotti says.
“He suggested we barcode the individuals by the number of notches in the top, middle, and lower section of the dorsal fin. By using this categorization system, we could manually match almost 5,000 photographs from 426 sharks in the first four years of study.”
Similar to how humans can be identified using fingerprints, tracking dorsal fins is a simple way to quickly identify sharks. The dorsal fin is the tall, triangular shaped fin that appears on the top of each shark; by observing notches on the fins, researchers can match the shark to one in the database.
By increasing our understanding and knowledge of white sharks, we will eventually maintain a healthier relationship with this ancient predator of the ocean. ~ Sara Andreotti
“White sharks are not as willing as people to provide their fingerprint, so snapping the right photograph takes some practice,” Andreotti says. “Unlike the movies, white sharks don’t swim with their dorsal fin out of the water at all times. Once the fin is completely out of the water, we need to take the photo as perpendicular as possible to the fin, so that the unique notches at the back become visible.”
Once the photos of the sharks are collected, each photo must also be cropped and edited so the notches appear as clear as possible since the sharks are being photographed against water.
The database software isolates the shark fin from the water in the background. Then, using an algorithmic technique called dynamic time-warping, Identifin automatically matches incoming photos to a shark image in the database, correctly matching images greater than 80 percent of the time.
Andreotti estimates that by using Identifin, the time it takes to complete her work is 1/10th of the original identification system, which was previously comprised of manually entering the photos.
Andreotti says her research offers important insights about how to protect white sharks going forward.
“The white shark is a top predator that has been demonized by human society for centuries,” Andreotti says. “The Identifin project shows how we can recognize individual sharks and monitor them long term. By increasing our understanding and knowledge of white sharks, we will eventually maintain a healthier relationship with this ancient predator of the ocean.”
Next up, Andreotti plans to adapt the Identifin database to aid other marine species identification efforts. Identifin will allow unprecedented insight into the movement of marine species in need of protection.
As in so many fields, computation is increasing the scope and abilities of conservation efforts.