Automated Detection of Rare and Endangered Anurans Using Robust and Reliable Detection Software
Date
2015-07
Authors
MacLaren, Andrew Rance
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Amphibian populations are experiencing rapid rates of decline, the causes of which are sometimes controversial. The vocalization of the male anuran is used as an indication of a potential breeding event. Researchers have been relying on these vocalizations to monitor the health, reproductive status, and diversity of anuran populations for centuries. As technology advances so does our ability to innovate and improve the way anuran populations are monitored. One such innovation comes in the form of portable commercially available audio recording devices (ARD). These tools enable researchers to capture the sounds produced by populations of any vocalizing animal species and analyze them using machine-learning techniques of pattern recognition. The application of these techniques is understudied and not well documented for anurans. I conducted rigorous testing of these techniques to improve methods of monitoring populations of the endangered Houston Toad (Bufo houstonensis). The desired result of these tests would be a reliable and robust tool for recognizing the call of the Houston Toad. This would allow researchers to search vast quantities of digital audio files for the unique sound of this animal. I also compared the efficacy of this machine-learning technique to a highly trained professional listening for the call. Researchers often doubt the reliability of automated techniques, thus my recognizer must perform capably. Additionally, I employed these automated machine-learning techniques to document the presence or absence of the Houston Toad in two counties of Texas, and then coupled those data with highly resolute details of the environmental conditions to examine calling activity of the Houston Toad and graphically visualize this behavior across a complete chorusing season.
Description
Keywords
Houston Toad, Automated detection
Citation
MacLaren, A. R. (2015). <i>Automated detection of rare and endangered anurans using robust and reliable detection software</i> (Unpublished thesis). Texas State University, San Marcos, Texas.