Date of Award

Spring 2025

Project Type

Thesis

Program or Major

Integrative Biology

Degree Name

Master of Science

First Advisor

Laura N Kloepper

Second Advisor

Megan A Cimino

Abstract

Adélie penguins (Pygoscelis adeliae) are bellwethers for Antarctic climate change because their breeding phenology is closely linked to sea-ice extent. Researchers need long-term, large-scale phenology data to make informed Antarctic climate and conservation insights. However, traditional breeding phenology data collection is resource intensive, limiting research to small colonies located near research stations. Passive acoustic monitoring (PAM) is a low-disturbance, easily scalable method which facilitates large-scale data collection and has been successfully applied at multiple seabird colonies to collect breeding phenology data. In this thesis, I provide a framework to implement a custom BirdNET wind detector to identify wind distorted recordings, a common problem with PAM data collection at seabird colonies. The wind model identified high wind with a 0.94 recall and 0.76 precision; a 0.91 minimum confidence score results in a 90% probability of a true high wind detection. I also demonstrate the viability of PAM to track discrete stages of Adélie breeding phenology. General linearized mixed models reveal significant relationships between acoustic indices—mathematical summations of energy distribution in a recording—and breeding stage. The Bioacoustic Index experienced slight, but significant, value increases across stages. The Acoustic Complexity Index and Root-mean-square Pressure (RMS) experienced increasingly higher values from incubation to guard, and guard to post-guard, and then rapidly declined during fledge (91% and 67% respectively). RMS best correlated to chick numbers during the breeding season (Pearson’s R = 0.81). This study represents the first step in building an acoustic workflow to passively monitor Adélie breeding phenology.

Download

Share

COinS