Application of autonomous recording units to monitor gobbling activity by wild turkey

Autor: Michael J. Chamberlain, L. Mike Conner, James A. Ruttinger, Mary M. Streich, Derek S. Colbert, Robert J. Warren
Rok vydání: 2015
Předmět:
Zdroj: Wildlife Society Bulletin. 39:757-763
ISSN: 1938-5463
DOI: 10.1002/wsb.577
Popis: Autonomous recording units (ARUs) allow simultaneous sampling at multiple locations and potentially offer an innovative way to improve assessments of gobbling chronology in male wild turkeys (Meleagris gallopavo spp.). We conducted research on 2 similar study sites in southwestern Georgia, USA, by deploying 14 ARUs across the 2 sites during 2011 and 2012. Autonomous recording units were useful at recording gobbling activity, allowing a single researcher to collect and analyze 19,957 hr of data in approximately 1,000 hr during 2 spring breeding seasons. Additionally, ARUs were capable of recording a gobble from a bird on the roost up to 207 m away; gobbles could be identified using autonomous call recognition (ACR) software. We used an abbreviated sampling regime to record gobbling activity, which displayed a similar pattern in gobbling activity when compared against continuous-daytime sampling. We used ACR software to develop a recognizer file that autonomously searched recordings and provided a list of candidate vocalizations that may represent turkey gobbles. Although we were able to develop a recognizer file that detected between 70.1% and 78.2% of all gobbling activity recorded, it also had false positive rates of 99.8–99.9%. The software performed poorly on account of inability to discriminate between turkey gobbles and other sounds (i.e., crow calls [Corvus spp.] and background noise). This resulted in lengthy data processing due to the large number of false positives that had to be evaluated individually. We recommend that future studies examine the efficacy of a variety of ACR software in identifying turkey gobbles. If this approach is not viable, we recommend future studies consider altering study design by focusing sampling efforts to specific times of day, thus reducing the number of audio samples recorded and subsequent processing time. Alternatively, researchers could either listen to the data or visually search the data for gobbling activity with spectrogram-viewing software. © 2015 The Wildlife Society.
Databáze: OpenAIRE