Telling one dolphin from another
Identifying individual dolphins using photos of the unique pigment patterns on their fins can help in the management of the species according to Niwa scientist Dr Krista Hupman.
Cetacean biologist Dr Hupman, lead author of two new scientific papers outlining an accurate method of counting dolphins by using photo identification, said individual animals’ dorsal fins had been photographed to establish a catalogue of dolphins that could be differentiated by differences in their markings and pigmentation patterns.
Scientists used catalogues to determine how many animals occupied a region and whether the population is in decline, but some dolphins could be hard to count as they occurred in large populations over vast areas, and may lack distinctive markings.
For those reasons many scientists shied away from using photo-identification to estimate populations, using aerial or shipboard surveys instead, methods Dr Hupman said may not be the best, and limited the ability to learn about individual animals.
There was a way to use photoidentification to estimate the number of poorlymarked dolphins, but it involved establishing new methods such as using dorsal fin pigmentation patterns for individual identification.
Over 2010-13 Dr Hupman and a team of researchers collected more than 240,000 images of dolphins in the Hauraki Gulf at different times of the year. Almost 31,000 of those photos were deemed of sufficient quality to use in the identification process.
They were compared manually, using nicks, notches and pigmentation patterns, resulting in 2083 individuals being identified.
“This study formed the largest catalogue of common dolphins worldwide, which is a significant achievement for everyone involved in this project,” she said.
“This is also the first study using photoidentification to determine the abundance of a large population of dolphins, and demonstrates that it is possible for species with [few] distinctive markings. The combination of dorsal fin edge markings and pigmentation patterns provided a robust method for individual identification.”
Identifying dolphins solely by the pigmentation on their dorsal fins had been thought to be unreliable, because the pigmentation changed over time and was not always visible, but it had been found on more than 95 per cent of adult common dolphins in the Hauraki Gulf. They also examined dolphins with photographic records spanning more than 10 years, and found no significant changes in the pigmentation pattern.
Now the team wanted to create a fully computer-automated system for identification whales and dolphins from photographs and video taken in the field.
The first step was applying an algorithm to classify individuals on pigmentation patterns alone.
“It used to take up to an hour to compare one dolphin image to the entire catalogue of individuals. However, by using these algorithms, we can match an image to the catalogue in a matter of seconds,” she added.