Between 365 and 988 million birds die from crashing into windows in the United States each year, according to a new report. That may be as much as 10 percent of the estimated total bird population of the country.

The estimate puts windows behind only cats as the largest source of human-related menaces that kill birds directly.

The biggest share of the collision deaths comes not from glass massacres at skyscrapers but from occasional collisions with the nation’s many small buildings, says Scott Loss of Oklahoma State University in Stillwater. “It’s death by a million nicks.”

Buildings four to 11 stories tall account for about 56 percent of deaths in the new estimate, Loss and his colleagues report in the Condor: Ornithological Applications. Residences that are one to three stories tall make up around 44 percent, with skyscrapers representing less than 1 percent.

Any given small building kills only a few birds each year, vs. the 24 expected to die annually at a single skyscraper. But the United States has about 15.1 million low-rises and 122.9 million small residences, and only about 21,000 skyscrapers. Loss applauds efforts to make skyscrapers bird-friendly, but he cautions that protecting birds takes a broader effort.

Some species appear especially vulnerable to the deceptions of windows, Loss and his colleagues find. Among the possible reasons are disorientation from artificial lights for birds on long-haul migrations at night. Compiling data from all kinds of buildings, the team found that Anna’s hummingbirds, black-throated blue warblers, ruby-throated hummingbirds, Townsend’s solitaires and golden-winged warblers topped the risk list.

There’s no nationwide reporting of birds’ thumping into glass or succumbing to a paw, so estimating death tolls has long been difficult and controversial. The new estimate of mortality from windows, based on statistical analysis of 23 local studies, comes close to an old estimate (100 million to 1 billion) that had been derided for its simple, back-of-the-envelope approach. “We were a little surprised,” Loss says.

There are plenty of uncertainties in extrapolating from small, diverse, local studies, particularly in trying to estimate overall species vulnerabilities, says Wayne Thogmartin of the U.S. Geological Survey in LaCrosse, Wis. But even such imperfect science has value, he says. For one thing, it may inspire people to start filling in gaps in data.

The total for window kills isn’t the whole story, though, says ornithologist Daniel Klem Jr. of Muhlenberg College in Allentown, Pa., who did the earlier calculation: “The moral imperative of preventing even one unwanted and unintended death of these utilitarian and aesthetically pleasing creatures is, or should be, compelling enough.”

— Science News