NCSU biologist finds reporting flaws in nation's bird counts

Every spring, dedicated bird-watchers armed with clipboards and binoculars embark on roadside counts to record how many birds they can see and hear. These long-running censuses, such as the North American Breeding Bird Survey, help scientists estimate trends in bird populations. Or maybe not.

New research by N.C. State University scientists suggests bird surveys that rely on identifying birds by their calls may be seriously flawed and could understate bird populations.

A team of researchers led by Theodore Simons, a biologist at N.C. State, found that trained bird-watchers aren't as good at hearing birds in the wild as previously thought, and their powers of perception drop sharply with even small increases in background noise. Factors such as traffic noise and observers' inaccurate spatial perceptions could inadvertently bias bird counts, they say.

"The real upshot of our work is not to question specific estimates for individual species," Simons said, adding, "We've probably been assuming our methods and data are better than they really are, and there are a lot of factors that can introduce bias into our results."

"In general, we were really surprised at how many mistakes people were making and how many birds were calling that observers were missing," he said.

To test observers' ability to identify birds by their calls, Simons and his fellow researchers developed Bird Radio, a series of about 45 remotely controlled playback devices that mimic birds singing in the woods. The playback devices, controlled by a laptop computer, were placed at distances of up 200 meters and heights of up to 15 meters from a central point in Howell Woods, an environmental learning center managed by Johnston County Community College and chosen because it's quiet.

Noise level matters

Quiet is important, because bird-watchers doing surveys are much more likely to identify birds by hearing them than by seeing them. Depending on the terrain, roughly 70 percent to 90 percent of birds are identified by ear.

"You can hear them from much farther away -- and hear them if you're not looking in the right direction," Simons said.

People participating in the experiment would stand at the designated spot and try to log on a data sheet as many of the half-dozen or so species as they could hear in three to 10 minutes. Calls of ovenbirds, Acadian flycatchers and various warblers sounded for about 20 seconds, sometimes a cacaphony of several songs at once. The listeners also tried to determine how many birds of a given species were present by estimating how far away each bird call was.

Simons said about 50 experienced birders took part in the experiments. The tests were conducted only in winter months from November 2005 to March 2007, when none of the migratory species used in the recordings would actually be present to bias the experiment.

The experiment showed even trained birders are limited in how well they can detect birds at different distances and how many birds they can hear at once.

"If you are trying to measure the distance to the sound source, your ability to map it accurately is much worse than we thought," Simons said. "Our ability to decide whether we hear two birds or one bird facing in two different directions is very limited."

The study found that even small amounts of background noise, from rustling leaves, wind or automobile traffic, led to a 40 percent decrease in the ability of observers to detect singing birds, Simons said.

"We know our world is getting noisier and noisier," Simons said. "If we do these survey routes over a 40-year period and counts are going down, does that mean the population is going down or our ability to detect these birds has gone down?"