Many hands, or in this case, many eyes, make light work. It’s a saying that scientists studying wildlife on the Serengeti recently put to the test.
Dr. Craig Packer, founder of the Serengeti Lion Project, and one of his graduate students, Ali Swanson, wanted to understand how so many species were able to co-exist with lions, which are, Packer explains, “pretty nasty to all the other carnivores.” So they decided to use camera traps. The hidden cameras captured a short sequence of photos when triggered by motion, creating a sort of who’s who at the watering hole.
In 2010, Swanson created a grid of 225 camera traps so that she could see how “different carnivores managed to avoid each other.” But she was quickly overwhelmed by the large number of photos each month. Enter the citizen scientists. Teaming with Zooniverse, Swanson and another graduate student, Margaret Kosmala, asked for help from laypeople to identify the animals photographed by the camera traps. And with that, the Snapshot Serengeti project was born.
Here’s how it worked: every four to six weeks from June 2010 to May 2013, field assistants visited the camera traps, replacing batteries and swapping out memory cards. The photos they collected were put onto hard drives and sent back to Minnesota, loaded onto a supercomputer at the University of Minnesota, and linked to Zooniverse where volunteers viewed the photos at random. The volunteers could view as many photos as they wanted, and their classifications were recorded.
As for quality control, Packer explains how it worked, “Our computers kept a running tally of everyone’s classifications for each photo. If the first five people all said that there was no animal visible in a particular photo, it was classified as “empty” and taken off line. If 10 people were unanimous in saying that an animal was, say, a giraffe or warthog, we accepted the consensus, and, again, the photo was retired. If the first 10 people showed any disagreement in their classifications, the photo was kept online until they were assessed by 25 different people, whereupon we accepted the plurality vote.”
And how did the citizens perform? “It turns out that, collectively, citizen scientists are extremely good in correctly identifying the species in these photos: compared to the judgements of a panel of experts who have viewed a selection of 4,400 photos, the Snapshot volunteers were correct about 97% of the time!”
After the photos were classified, the date and location of each image was used to place the animals across the grid. “We can see how each species has moved around a 700 square mile area month by month for the past three years,” says Packer.
The numbers are impressive. In all, 32,935 registered volunteers (and almost as many unregistered users) classified 1.54 million photographic sequences. (When triggered, the cameras shot three frames during the day and one frame at night.) It’s obvious that the arrangement benefited scientists. With numbers that large, Packer says it would have taken scientists months or years to finish what took citizens a couple of days.
But what was the draw for the volunteers? Not only did they have a front row seat on a safari from their living room, but there was an element of discovery in looking through the photos. “Nothing, nothing, wildebeest, nothing, wildebeest, nothing nothing, then LEOPARD!! It’s like hitting the jackpot!” Packer says. “We also had an active blog and chat-rooms for volunteers, so a lot of people felt like part of the research team.”
These covert observations have opened up a world of information about community dynamics.“We’ve seen a lot of interactions between species that I doubt anyone has ever observed before, like bat-eared foxes chasing off an aardwolf, and topi saying hello to warthog.” Some of Packer’s favorites? “Mating porcupines, oxpeckers roosting in the crotch of a giraffe, a small herd of eland posed like a bouquet, spectacular views of zebra and wildebeest herds running directly towards the camera, and any LEOPARD!” He really likes leopards.
The interview with Dr. Craig Packer was conducted by Tess Vincent.