Skip to main content

Female choice scores and Peak Frequency and Duration in calls from Wood frog chorus recordings

Cite this dataset

Calsbeek, Ryan (2022). Female choice scores and Peak Frequency and Duration in calls from Wood frog chorus recordings [Dataset]. Dryad.


A limitation in bioacoustic studies has been the inability to differentiate individual sonic contributions from group-level dynamics. We present a novel application of acoustic-camera technology to investigate how individual wood frogs calls influence chorus properties, and how variation influences mating opportunities. We recorded mating calls and used playback trials to gauge preference for different chorus types in the laboratory. Males and females preferred chorus playbacks with low variance in dominant frequency. Females preferred choruses with low mean peak frequency. Field studies revealed more egg masses laid in ponds where males chorused with low variance in dominant frequency. We also noted a trend towards more egg masses laid in ponds where males called with low mean frequency. Nearest neighbor distances influenced call timing (neighbors called in succession) and distances increased with variance in chorus frequency. Results highlight the potential fitness implications of individual-level contributions to a bioacoustic signal produced by groups.


To evaluate female choice between choruses, we combined recordings of individual males to construct three chorus types: 1) LFLV choruses had low peak frequency (e.g., PF ~1542 Hz) with low variation in peak frequency (measured as the interval above and below PF before the amplitude dropped 20dB below the peak e.g., 1329-1937 Hz). 2) HFLV choruses had high peak frequency (e.g., 1808 Hz) with low variation (e.g., ±20 dB range= 1413-2037 Hz) and 3) HV choruses had both high and low frequency calls (PF 1807 and 1542 Hz) and thus high variance in call frequency. Two versions of each chorus-type were produced and were used in alternating order between trials. Each frog heard the HFLV and LFLV choruses, LFLV and HV choruses, or HFLV and HV choruses from two of four speakers. Choruses were played through alternate pairs of speakers in each consecutive trial.  The other two speakers served as silent controls. Trials lasted ten minutes or until the focal frog had moved to within 5cm of a speaker and remained at that speaker for at least one minute. Trials in which the focal frog did not move towards a chorus or did not make a definitive choice (14, 10, and 13 trials in the three treatments respectively) were recorded ‘NA’.

We recorded natural choruses at each of 11 ponds using an acoustic camera mounted to a tripod. All recordings were made between 8:00 a.m. and 1:00 p.m.  Acoustic cameras integrate a digital video-camera with a microphone array to map sound onto acoustic still or video images. Frog choruses were recorded using a Ring48 AC Pro Polytech acoustic camera. This configuration consisted of a 0.75-meter rigid carbon fiber ring with 48 calibrated microphones (+/- 0.5 dB sensitivity) and a centrally located video camera. Data were aggregated in a Polytech Multi Channel Data Recorder (mcdRec). Data were extracted using the NoiseImage software v. 4.11 (Gfai tech). The NoiseImage software uses time of arrival differences among microphones to reconstruct sound sources on the video of the scene, producing a video with an overlaid heatmap of sound sources. All videos were reconstructed with a framerate of 25 frames/second and an overlap of 3 frames. This process resulted in one WAV file per frog that contained all calls of that individual. To extract acoustic measurements from these recordings, we compiled all of the WAV files into a multichannel file, with one channel per frog. We then converted the video frame numbers into time stamps using the frame rate of the video and used a custom script to build a selection table for the Raven Pro sound analysis software. To maximize the precision of the measurements, we used two different sets of spectrogram parameters. We exported peak frequency values using an FFT (Hanning) size of 4096 samples to facilitate frequency resolution, giving a 3 dB filter bandwidth of 33.7 Hz. We exported 90% duration values using an FFT of 256 samples, resulting in FFT bins of 0.0027 sec. All spectrograms were computed with 50% overlap.

Usage notes

Missing values are coded NA

Readme file contains all relevant information



National Science Foundation, Award: DEB-1655092