2020 Volume 26 Issue 1 Pages 1-9
Passive acoustic monitoring is an effective method to observe feeding behavior of dugongs in their natural environment. In this paper, we developed a new method to automatically detect the feeding sounds of dugongs from acoustic recording data. In January 2018, we deployed an autonomous underwater sound recorder on the seafloor in an intertidal seagrass bed off Talibong Island, Trang, Thailand. We developed custom-made software to detect the feeding sounds of dugongs automatically. The feeding sound consists of a set of pulses with a constant interval. A morphological opening filter was applied to identify pulses and autocorrelation coefficient threshold was used to extract the feeding sounds. Typical ambient noise that caused false detections included the sounds of snapping shrimps, other crustaceans, rainfall, waves, and boat traffic. Applying a threshold on water depth effectively reduced the false detection rate from 16.4 to 4.4%, and the correct detection rate increased from 72.2 to 76.8%. The proposed method will reduce time and effort for manual detections, and thus contribute to extension of temporal and spatial scale of passive acoustic monitoring of dugong feeding behavior.
