Ecologists Bested by Drone in a Competition to Count Birds

A novel experiment pitted a drone against experienced wildlife spotters to estimate the size of faux flocks. Despite the results, some ornithologists are wary of the new technology.

Two-thousand rubber duckies would make for a pretty epic bath. Or in this case, a first-of-its-kind science experiment. Ecologist Jarrod Hodgson used thousands of rubber duck decoys in a new competitive experiment that matched experienced ecologists against a drone to see which counts wildlife more accurately.

Hodgson wasn鈥檛 merely entertaining himself, though he admits the experiment was great fun. Instead, he was dead set on answering a question central to modern conservation: How can scientists best estimate the size of wild animal populations? This information is critical; indeed, scientists use population estimates to decide whether a species is in trouble鈥攊ncluding listing it as 鈥渢hreatened鈥 or 鈥渆ndangered鈥濃攁nd, ultimately, whether it needs urgent attention and funding to bring its kind back from the brink.

While it might be simple in theory, in practice counting a large troop of animals is a serious challenge. 鈥淧eople working in conservation are constantly struggling to come up with population figures,鈥 says Kenn Kaufman, a renowned bird expert and field editor at 探花精选. Ecologists in the field will typically try to find a good vantage point from which to observe a flock, and then use rudimentary technologies鈥攂inoculars, spotting scopes, and hand-held tally counters鈥攖o estimate its size.

It鈥檚 often easier to get an accurate count from an airplane, sometimes with the help of aerial photos, but this approach costs time and money, says Andrew Pierson, director of conservation at the at Rowe Sanctuary in Nebraska. During migration, the center's conservation partners on the Platte River estimate how many Sandhill Cranes travel through the area鈥攁nd it isn鈥檛 easy. 鈥淚t takes a pilot, hiring a plane, a staffer or two to do the count, and another staffer to sit in the office and analyze the results,鈥 he says. And some ecologists worry that the loud plane engines disturb the wildlife they鈥檙e aiming to assess and protect.

In comparison, remotely piloted aircrafts, or drones, seem to strike a middle ground. Airborne, they can get a better view than a terrestrial observer. They can capture photos for analysis after researchers have left the field. And unlike airplanes, drones are relatively small and quiet, reducing the risk of spooking the birds.

As a result, 鈥渢here鈥檚 been an increasing use of drones among ecologists and environmental managers,鈥 Hodgson says. 鈥淭he proliferation has been phenomenal.鈥 (The Federal Aviation Administration estimates that the commercial drone fleet will increase in the U.S. between 2016 and 2021). Drones have have been used to monitor orangutans in Indonesia, elephants in Burkina Faso, and, increasingly, they鈥檝e been used to count birds.

But are drones really superior to traditional observers? 鈥淭here鈥檚 a growing idea that they are better,鈥 Hodgson says, but 鈥渢here isn鈥檛 much actual statistical evidence.鈥

So he decided to put drones to the test. In 2016, Hodgson, a PhD candidate at the University of Adelaide, arranged 2,000 rubber ducks into 10 colonies on a public beach in South Australia in a nesting pattern modeled after the Greater Crested Tern. Like the proctor of a bubblegum-counting contest, he knew the true size of each colony鈥攁nd then he put two teams head-to-head to compete for the most accurate count. On one team were experienced wildlife spotters, equipped with binoculars or scopes, stationed about 120 feet away and at roughly the same altitude as the faux colonies. On the other was an off-the-shelf quadcopter, armed with a digital camera, and a handful of untried volunteers who reviewed each image manually, tallying the number of birds pictured.

When the results came in, Team Drone was the undisputed winner. On average across all colonies, the drone鈥檚 estimates were significantly closer to the true number of ducks than those made by ecologists; in some cases, they were 96 percent more accurate, depending on the drone鈥檚 flying height. And when the researchers discarded a subset of drone images that came out blurry, the lead was more pronounced: The drone counts were up to 98 percent (and no lower than 92 percent) more accurate than the human counts. The study was published in Methods in Ecology and Evolution.

To cut down on time, Hodgson had another idea. 鈥淲e wanted to see if we could automate the count and still be accurate.鈥 The researchers devised an algorithm to process drone photos semi-automatically, and its accuracy was nearly identical to the manual analysis.

Even though the study's results were overwhelmingly in favor of drones, some ornithologists hesitate to embrace the new technology.

鈥淚t鈥檚 another form of recreation that might have an impact鈥 on birds, says Larry Niles, a wildlife biologist who has been doing surveys for 40 years. 鈥淲ith anything like this, people will be using them all over, so now we have a new source of disturbance.鈥 Pierson echoed his concern.

Hodgson is aware that disturbance is a risk. In fact, he guidelines on how to fly drones to minimize disturbance to wildlife in 2016. But he himself acknowledges that there鈥檚 still a lot we don鈥檛 know. 鈥淭here鈥檚 basically no published data on birds鈥 physiological responses,鈥 he says. When a drone approaches a population of birds, for example, 鈥渢hey might just stick there and look fine鈥 even if they鈥檙e experiencing stress responses. Currently, Hodgson is trying to quantify those responses by monitoring how the heart rate of birds changes when a drone is nearby. So far, he says, the data are inconclusive.

The ornithologists interviewed aren鈥檛 concerned only with the birds鈥 health, but also with that of the birding and scientific community. 鈥淏ird surveys should belong to people who love birds,鈥 Niles says. He worries that eliminating people, including passionate volunteers, from surveys will reduce support for conservation measures. 鈥淲e鈥檙e constantly trying to professionalize our survey when in fact we have legions of people willing to take part,鈥 he says. 鈥淲hether or not we use drones, surveys should involve people.鈥

Concerns aside, none can deny that drones will have a place in conservation. 鈥淲e鈥檙e seeing a shift in that direction,鈥 Pierson says. 鈥淲e have a small drone that we use for various things already.鈥 The key is in how drones are used鈥斺渨ith the right application, [they] could be really effective,鈥 Kaufman says鈥攁nd whether they鈥檙e used responsibly. That鈥檚 why Hodgson鈥檚 work, including all those rubber ducks, is so necessary.