Wildfires are a natural part of forest lifecycles in the western United States, helping recycle nutrients, diversify forest structure, and even germinate seeds. However, climate change and a history of post-colonial fire suppression has resulted in increasingly large and destructive fires. To make forests more resilient to these destructive events, forest managers are increasingly turning to prescribed fires.
Since 2017, a team from the California Academy of Sciences, University of California, and the U. S. Forest Service have been measuring avian biodiversity in the Caples Creek watershed using both traditional point counts and Autonomous Recording Units (ARUs). This watershed was burned with a prescribed fire in the fall of 2019, which allows us to compare how the species composition changes following a prescribed burn.
In 2020, I joined this team as a National Science Foundation REU intern with the California Academy of Sciences Summer Systematic Institute. Through this program, I conducted research designed to compare the effectivity of ARU recordings paired with a machine learning program against human point counts. I compared 1.7-million machine-identified sound events with human-identified manual annotations for the same sound files and found mixed success in the ability of our machine learning program to accurately identify the bird community.
I presented my findings from this research at the 2020 Summer Systematic Institute Research Symposium.
The U.S. Forest Service is also working to expand this research to study how the carnivore community is responding to the prescribed burn. In the fall of 2020, I installed baited game cameras across the Caples watershed and monitored them on a weekly cycle. This research is likewise ongoing as the forest responds to the prescribe burn. Stay tuned!
Conducting a point count
Caples Crew 2020
Sierra Nevada mountains
Sierra Nevada mountains
Installed game camera
Wildfire smoke
