Timothy C. Mullet ^1* Almo Farina ² John M. Morton ³ Sara R. Wilhelm ¹

• ¹ US National Park Service, Seward, United States
• ² Department of Pure and Applied Sciences, School of Conservation and Restoration, University of Urbino Carlo Bo, Urbino, Italy
• ³ Alaska Wildlife Alliance, Solodotna, Alaska, United States

Given that ecosystems are composed of sounds created by geophysical events (e.g., wind, rain), animal behaviors (e.g., dawn songbird chorus), and human activities (e.g., tourism) that depend on seasonal climate conditions, the phenological patterns of a soundscape could be coupled with long-term weather station data as a complimentary ecological indicator of climate change. We tested whether the seasonality of the soundscape coincided with common weather variables used to monitor climate. We recorded ambient sounds hourly for five minutes (01 January to 30 June) over three years (2019 -2021) near a weather station in a subarctic ecosystem in south-central Alaska. We quantified sonic information using the Acoustic Complexity Index (ACItf) and used machine learning (TreeNet) to find relationships between ACItf and weather data. We grouped ACItf according to the timing of prominent seasonal events (e.g., daily temperatures >0 °C, no snow cover; green up, dawn biophony, road-based tourism) and identified distinct sonic phenophases (sonophases) for groups with non-overlapping 95% confidence intervals. In general, sonic activity increased dramatically as winter transitioned to spring and summer. We identified two winter sonophases, a spring sonophase, and a summer sonophase, each coinciding with hours of daylight, temperature, snow cover, and the prevalence of animal and human activities. We discuss how monitoring sonophases in conjunction with weather data can be useful as a holistic approach to understanding the ecological effects of climate change in subarctic environments.