Birdsongs indicate population size’

Ecologists have at last worked out a way of using recordings of birdsong to accurately measure the size of bird populations.

Developed by a joint team from the US Geological Survey and University of Otago, the technique is an innovative combination of sound recording with spatially explicit capture-recapture, a new version of one of ecologists’ oldest tools for monitoring animal populations.

Birds communicate by singing or calling, and biologists have long counted these cues to get an index of bird abundance. But it is much harder to work out the actual density of a bird population because existing methods need observers to measure either the distance to each bird, or whether they are within a set distance from the observer.

This is straightforward if birds are seen, but difficult when birds are heard but not seen.

Team leader Dr Murray Efford said a way has been found to estimate population density of birds or other animals that vocalise by combining sound information from several microphones.

A sound spreading through a forest or other habitat leaves a ‘footprint’. The size of the footprint depends on how quickly the sound attenuates.

Mathematically, there is a unique combination of population density and attenuation rate that best matches the number and ‘size’ of the recorded sounds. Computer methods were used to find the best match and thereby estimate density.

The researchers developed the method by recording the ovenbird — a warbler more often heard than seen — in deciduous forest at the Patuxent Research Refuge near Laurel, Maryland, USA. They rigged up four microphones close to the ground in a square with 21 metre-long sides. Over five days, they moved the microphones to 75 different points across their study area and recorded ovenbirds singing.

“Sound intensity and other characteristics can be measured from the spectrogram — the graph of the sounds — to improve density estimates. Archiving the sounds also makes it possible to re-examine them, or to extract additional information as analytical methods evolve,” he said.

The findings have been published in the ‘Journal of Applied Ecology’