Observer-free experimental evaluation of habitat and distance effects on the detection of anuran and bird vocalizations.

Acoustic surveys of vocalizing animals are conducted to determine density, distribution, and diversity. Acoustic surveys are traditionally performed by human listeners, but automated recording devices (ARD) are becoming increasingly popular. Signal strength decays, or attenuates, with increasing distance between source and receiver and some habitat types may differentially increase attenuation beyond the effects of distance alone. These combined effects are rarely accounted for in acoustic monitoring programs. We evaluated the performance of three playback devices and three ARD models using the calls of six anurans, six birds, and four pure tones. Based on these evaluations, we determined the optimal playback and recording devices. Using these optimal devices, we broadcast and recorded vocalizations in five habitat types along 1,000 m transects. We used generalized linear models to test for effects of habitat, distance, species, environmental, and landscape variables. We predicted detection probabilities for each vocalization, in each habitat type, from 0 to 1,000 m. Among playback devices, only a remote predator caller simulated vocalizations consistently. Differences of ~10 dB were observed among ARDs. For all species, we found differences in detectability between open and closed canopy habitats. We observed large differences in predicted detection probability among species in each habitat type, as well as along 1,000 m transects. Increases in temperature, barometric pressure, and wind speed significantly decreased detection probability. However, aside from differences among species, habitat, and distance, topography impeding a line-of-sight between sound source and receiver had the greatest negative influence on detections. Our results suggest researchers should model the effects of habitat, distance, and frequency on detection probability when performing acoustic surveys. To optimize survey design, we recommend pilot measurements among varying habitats.