People power
Ordinary members of the public are helping to study everything from puffins to penguins. But is their data any good – and can becoming a citizen scientist really save a species? Louise Tickle investigates.
How citizen scientists are helping researchers
On clifftops across Britain and Ireland last spring and summer, an army of puffin fans clicked their cameras as thousands of these much-loved seabirds flew back to their burrows, silvery fish clutched in glowing orange, yellow and blue beaks. Co-opted as citizen scientists by the RSPB as part of an initiative called Project Puffin, the 602 puffin enthusiasts – dubbed the ‘puffarazzi’ – then uploaded 1,402 photographs to the project’s website, together with an explanation of where and when each had been taken.
The images were scrutinised by a team of six specially trained citizen scientists led by RSPB seabird specialist Ellie Owen, in an effort to identify the species and size of fish each bird was bringing back to its nest. Puffins, Owen explains, are in serious trouble in many former strongholds. On the ‘Red List’ – indicating that a species has hit the highest level of conservation
concern – this charismatic auk is now one of the RSPB’s top research priorities.
“We had long been thinking about seabird diet and needed more data,” Owen says. “We know food supply is a major bottleneck for seabirds when they go into decline, as puffins are right now.” So she had the idea of creating the first national map of what puffins are feeding their chicks around the country, “and then relating that to the broader aim of working out why puffins in some colonies are finding food more easily than others.”
Owen decided to harness the love of puffins felt by so many visitors to coastal reserves, and crowdsource pictures of individual birds returning home from fishing trips. “If we’d paid researchers to get those photographs it would have cost hundreds of thousands of pounds, and we could never have afforded to do it,” she says.
Citizen science – drawing on the time, passion, skills and sometimes the technical expertise of ordinary members of the public – has become increasingly popular with researchers across a range of scientific disciplines over the past decade. Usually focused on the mass gathering or analysis of large data sets, it has been heralded as an innovative and, importantly, inclusive way to better compile and understand information about the world we live in. B
ut really, given that citizen science typically depends on cohorts of untrained volunteers, how useful are the findings of such research? “I think the answer is, it depends,” says Tim Birkhead, a professor of ornithology at Sheffield University. “The crucial thing is quality. Looking at broad general trends by looking at large data sets can be useful, if your methodology is robust and consistent.”
Birkhead is pleased that people increasingly want to feel involved in science, but warns that, “there’s no point in doing it unless it’s reliable. There’s a real risk that citizen science projects end up, crudely, as job creation schemes, keeping the public happy and entertained.”
Consistency is key. “If the methodology changes – and I see that kind of thing happening all the time on local nature reserves, with people coming along and saying “I’ve got a better method than you” – then you’re on a sticky slope,” Birkhead says. By contrast, he cites the swallow migration tracking platform on the eBird website (www.ebird.org) as an excellent example of the power of citizen science. “It offers superb visualisation, and if done over successive years is a valuable research tool.”
“CITIZEN SCIENCE HAS BECOME INCREASINGLY POPULAR ACROSS A RANGE OF SCIENTIFIC DISCIPLINES OVER THE PAST DECADE.”
Robust project design is a vital first step to gathering reliable data, agrees Oxford University’s professor Chris Lintott. He is a co-founder of Zooniverse, an online platform that hosts a huge variety of naturebased citizen-science projects. But he also notes that volunteers can do some jobs better than the experts. L
intott, an astronomer, stumbled into citizen science 11 years ago, when he realised he had “too many galaxies” for PhD student Kevin Schawinski to analyse. “In desperation, after Kevin had looked at 50,000 galaxy pictures, we put them up online and invited people to search for planets,” he explains. “On the first day, we were getting 70,000 classifications an hour. We soon realised that not only were people’s classifications better than machine recognition, they were also better than Kevin!”
There are ways to strengthen confidence that volunteer classifications can be relied upon. A case in point is Penguin Watch, which launched online in 2014 and now has almost 50,000 registered contributors. To date they have made over six million categorisations of time-lapse photos of penguins, chicks and eggs.
Penguinologist Fiona Jones explains how its researchers control for quality. “If any volunteer indicates that animals are present in an image, that image is next shown to a total of 10 people,” she says. “Each click they make on the photo gives us an x/y co-ordinate. By clustering the co-ordinates generated by the clicks of all 10 volunteers, researchers can then produce ‘average clicks’, each of which should represent one penguin adult – or chick, or egg.”
The point of Penguin Watch, Jones explains, is to understand the changing population dynamics of penguin colonies in the Antarctic. “Many colonies go completely unmonitored,” she says. Some may not even be known about: in 2014, a vast colony of Adélie penguins was spotted on the Danger Islands in the Weddell Sea, which had gone undiscovered until then. “Colonies could be in rapid decline and we’d have no idea,” Jones continues. “Since penguins are often thought of as a bellwether of climate change, it’s important to know what’s going on.”
Birkhead, however, warns that using untrained volunteers to collect data is risky. He adds: “While I don’t want to take anything away from the enthusiasm of the public, researchers have to make sure that their instructions are bombproof.”
But scientists can build in safeguards. Ensuring that untrained members of the public are asked only to carry out a simple, well-explained task, is critical to success, says PhD student Anabelle Cardoso of Oxford University, who set up Elephant Expedition to analyse the numbers of forest elephants – and other species – appearing in over two million camera-trap photos from her study site in Gabon. “If we were to ask volunteers to do more complicated tasks with more opportunity for mistakes, we would raise the number of times each image needs to be classified in line with the difficulty of the task,” she says. T
he Penguin Watch website takes volunteers through a swift, clear tutorial and straight into classifying adult penguins, chicks and eggs. The instructions couldn’t be simpler. Even so, it can be hard in half-light to distinguish between an adult and a fullgrown chick. Jones explains that if a bird is annotated by only one volunteer, it was probably done in error and can be discarded. But, she says, analyses of the project’s crowdsourced data have found that the public are “excellent” penguin spotters. “In fact, our volunteers are so diligent that they can spot things the experts miss!”
Identification has been an issue for the Shark Trust’s Great Eggcase Hunt, as the subtle differences between shark and
skate eggcases found on beaches can make identifying them tricky, even with the help of a detailed guide on the charity’s website. The solution has been a smartphone app, launched six years ago. Cat Gordon, the Shark Trust’s conservation officer, says that it enables volunteers to submit eggcase photographs instantly from the beach, rather than having to upload images manually from a computer at home.
“We’re finding that 80 per cent of records sent using the app come with an image, compared to 40 per cent of those sent by other methods,” Gordon says. This makes checking a volunteer’s eggcase identification much easier. “If we receive a record without an image and don’t already know the person is good at identification, then the data is approved but not verified, and we show the split on our results map,” she explains. “Some we will follow up on because we know we shouldn’t be getting that species in that place.”
It’s also vital, says Birkhead, to build double-blind checks into the project methodology. As an example of this not happening, he cites the RSPB’s hugely popular Big Garden Birdwatch. “You’re asking people to record the numbers of birds they see in an hour from their front room. But if, unbeknown to them, you set up video cameras beforehand and then compare the results, I’ll tell you what you’ll find. Everybody exaggerates!” B
irkhead laughs wryly. “It’ll be things like: ‘Oh, I saw a bullfinch a minute before my hour officially started, but I’ll include it anyway.’ So unless you build in careful quality checks you risk compromising your data.” Another danger of citizen science, observes Owen, is that you get “huge amounts of data but nothing that’s useable”. She was initially worried about getting hundreds of blurred puffin images that would be of no help at all in working out the variation in their suppers.
Careful thinking should also be done in advance of designing a project, “so you head off any biases early on,” Owen says. “For instance, we needed people not just to send through flashy pictures of puffins carrying big fish – we needed to be able to see whatever they’d managed to catch.” It seems the instructions worked: Owen has been able to use 89 per cent of the pictures sent in.
But what about untrained volunteers misidentifying species, so skewing the data? It’s easy enough, perhaps, to recognise a forest elephant, but different species of ladybird, for instance, might easily confuse a lay person. Given this risk, doing due diligence means having an expert check at least a proportion of the crowdsourced results against their own, skilled assessment of what is shown in an image.
At the Centre for Ecology and Hydrology, professor Helen Roy says they go even further. Roy pioneered online outreach to citizen scientists when she put the Ladybird Survey online in 2005. “For every record we get in, we’ll check it.” And with the launch of the Ladybird Survey app, the quality as well
as quantity of data she receives has risen as volunteers are able to take a photo of the ladybird they’ve found, as well as the exact time and location of their recording being automatically logged. R
oy points out that one of the joys of citizen science is the opportunity for researchers to be in regular touch with wildlife lovers who share their passion for a particular species. She has adopted social media with enthusiasm, and also messages volunteers individually to tell them more about any interesting species they record. When she needs to reclassify a misidentification, she contacts the volunteer to explain why.
“It feels like a community project, like a survey that belongs to everyone,” Roy observes. “Some people provide just one or two records, others survey their local woodland year-on-year, so it’s very much distributed – not in any way that’s formal, but in a beautiful maverick way. I love hearing people’s stories, and they love telling their stories about the experience of being around ladybirds.”
Outreach is, for some scientists, a fundamental driver for crowdsourcing information from the public. While the Shark Trust has compiled 170,000 valuable records over the past 14 years, Gordon says that research was not the initial motivator for the project, which was in fact designed “to show that we have sharks, skates and rays in our waters,” she explains. “People are still surprised!”
For Anabelle Cardoso, assessing the success of a citizen science project derives not only from whether the research has fulfilled its scientific aims but also from the quality of its relationship with volunteers. “Did they feel they gained enjoyment and knowledge from the process of engaging with it? Would volunteers educate those around them about the research?” she says.
Taking part might give volunteers a nice warm feeling, but can becoming a citizen scientist truly help to save a species? Volunteers might need to think long haul rather than sprint, but Roy is in no doubt that their efforts make a difference. “The value is just absolutely unimaginable,” she says. “The government, every year, produces its indicators, and many rely on volunteer generated data, so for communication to ministers and to inform their decisions on possible changes to land management, that policy link is very valuable.”
“It’s teaching us about puffins and their food sources, but the way people engage is also teaching us about citizen science methods,” says Owen. “The purpose is to be able to make better conservation plans for puffins. The reason we can’t do it at the moment is that while they may seem very well known and well loved, they’re one of the seabirds we know least about.”