Categorizing click trains to increase taxonomic precision in echolocation click loggers.

Area of Science:

• Marine biology
• Bioacoustics
• Wildlife monitoring

Background:

• Passive acoustic monitoring is crucial for studying marine mammals.
• Identifying delphinid species using C-POD data is challenging with current software.
• Echolocation click analysis is key to understanding cetacean behavior.

Purpose of the Study:

• To develop a method for distinguishing delphinid species based on C-POD data.
• To improve the accuracy and proportion of categorized echolocation click trains.
• To validate acoustic species identification with visual sighting records.

Main Methods:

• Compared C-POD click train features with continuous recorder frequency spectra.
• Utilized a generalized additive model to categorize click trains (broadband, frequency-banded, unknown).
• Pooled model predictions within acoustic encounters and applied a likelihood ratio threshold for categorization.

Main Results:

• Initial model achieved low categorization rates (30%) despite high accuracy for categorized clicks (0.02 error).
• Pooling predictions and using a likelihood ratio threshold increased categorization to 98%.
• Acoustically predicted species distribution aligned well with visual sighting data across 30 sites.

Conclusions:

• The developed method significantly enhances the ability to identify delphinid species from C-POD data.
• Improved categorization rates increase the utility of passive acoustic monitoring for marine mammal research.
• This approach offers a robust tool for mapping species distribution and informing conservation efforts.