Comparison of audio recording system performance for detecting and monitoring songbirds. Issue 1 (21st February 2013)

• Record Type: Journal Article
• Authors: Rempel, Robert S.; Francis, Charles M.; Robinson, Jeffrey N.; Campbell, Margaret
• Is Part Of: Journal of field ornithology. Volume 84:Issue 1(2013)
• Page Start: 86
• Page End: 97
• Publication Date: 2013-02-21
• Subjects: Birds -- Periodicals; Ornithology -- Periodicals; Ornithologie -- Périodiques; Oiseaux -- Baguage -- Périodiques; Oiseaux -- Périodiques; 598
• DOI: 10.1111/jofo.12008
• Languages: English
• ISSNs: 0273-8570

Abstract

<abstract abstract-type="main" xml:lang="en"> <title> <x xml:space="preserve">Abstract</x> </title> <p> <bold>ABSTRACT</bold> Acoustic recording systems are being used more frequently to estimate habitat occupancy or relative abundance, and to monitor population trends over time. A potential concern with digital recording systems is that changes in technology could affect detectability of birds and cause bias in trend estimates based on counts of birds detected. We evaluated several currently available commercial recording systems ranging from low‐cost multipurpose digital recorders to custom‐designed wildlife recorders (US$250–$7000 price range) to examine possible differences among systems in species detection. We made recordings during Breeding Bird Surveys (BBS) counts using several units concurrently, and asked several expert birders to listen to the recordings in a factorial design. We found that birders detected, on average, 10% fewer species on some units compared to others, though there was high variance. Analysis of a subset of recordings, using spectrograms and repeated listening, suggested that ∼90% of species on each BBS stop could be clearly detected on all units. The remaining species could be identified on at least one unit, but were hard or impossible to detect on others. We found that the recording unit with the lowest empirical signal‐to‐noise‐ratio (SNR) had the lowest number of birds detected on the BBS recordings, and that frequency‐specific SNR differed among units. Missed detections were likely related to variation in internal noise and frequency‐dependent sensitivity of the units, and were an issue for all systems regardless of price. We caution that researchers using recorders need to consider variation among recording systems in their study design, particularly for long‐term monitoring programs.</p> </abstract>