Deploying acoustic recorders to study songbirds in decline

NFWF has grown to become the nation’s largest private conservation grant-maker in part due to its sharp focus on measurable outcomes for wildlife.

The Foundation closely monitors project results, through both internal and third-party analyses, to ensure that precious conservation funds are invested in the most effective conservation efforts.

So how does one measure real-world effects of conservation projects, especially when the critters being studied come and go as they please? Take, for example, golden-winged and cerulean warblers. These songbirds spend their summers in the Appalachian region and their winters in Central and South America. How can you tell which conservation practices generate the greatest benefits? Do you enlist an army of birders to stake out the woods, listening and watching every day?

Ornithologists have deployed a variety of techniques to monitor bird populations. Point counts, nest counts, netting, banding, tracking, observation — all are valid and time-honored tools of the trade. But now, thanks to the work of a cadre of innovative ornithologists, we can, in essence, deploy that theoretical army of birders to keep track of the birds — all day, every day, for months or even years.

In 2022, NFWF invested more than $356,000 into efforts by a research team at the University of Pittsburgh to deploy and train people to use hundreds of automated acoustic recorders in blocks of forest being restored in Maryland, Pennsylvania and West Virginia. These low-cost acoustic loggers, about the size of a deck of cards, can record bird calls and songs continuously.

Artificial intelligence, along with human listeners, will analyze thousands upon thousands of hours of recordings. Sophisticated software systems called “automated machine learning species classifiers” can analyze the recordings and determine the presence or absence of golden-winged warblers, wood thrushes, cerulean warblers and at least four additional bird species in each audio file.