By Susan Enright.

Driven by a desire to see endangered birds return from the brink of extinction, and readied with a bachelor of science in biology degree from the University of Wisconsin-Madison, Amanda Navine arrived at the University of Hawaiʻi at Hilo in 2019 to earn a graduate degree in tropical conservation biology and environmental science.

“I knew I wanted to work in avian conservation, and there are few places on Earth where birds have greater need for passionate conservation biologists than Hawaiʻi,” says Navine. “I had never been to Hawaiʻi before, but I quickly learned while researching the university that the native ecosystem is perilously close to collapse and I felt I had something to contribute to the fight against extinctions.”

Navine says from her very first conversation with Patrick Hart, UH Hilo professor of biology who specializes in forest birds, she knew they would work well together. When he offered her work in his lab, she was delighted to accept.

Professor Hart’s lab, the Listening Observatory for Hawaiian Ecosystems Bioacoustics Lab (commonly called LOHE Lab), is a center for research on the ecology and conservation of native Hawaiian forests and birds. Using bioacoustic technology, Hart and his lab teams routinely conduct studies on how birds communicate with each other through sounds and also how scientists can use those sounds to improve monitoring the birds’ distribution and abundance over time.

The lab was the perfect fit for Navine. She worked and conducted her research there, and following graduation moved seamlessly into her career as the lab’s acoustic bioinformatics specialist.

Graduate school at UH Hilo

Navine first came to Hilo in the spring of 2019, the semester before she began graduate school, to assist in a project using bioacoustics to assess the impact of Rapid ʻŌhiʻa Death on bird communities.

“We deployed AudioMoths (very tiny recorders) in infected and healthy forests to determine if the avian species assemblages were becoming non-native species dominant in forests heavily altered by ROD,” she explains.

As a graduate student, her research was centered around gut microbiomes and determining if the Hawaiʻi ʻamakihi is more resistant to avian malaria, the primary disease threat to native honeycreepers, because their gut harbors specific microbes.

“We collected fecal samples from ʻamakihi and warbling white-eyes, who are immune to malaria mortality, to assess similarities and identify candidates for microbes with antimalarial properties,” says Navine. Her work is published here:

• Microbiomes associated with avian malaria survival differ between susceptible Hawaiian honeycreepers and sympatric malaria-resistant introduced birds (in the journal Molecular Ecology, Oct. 2022)

While a student, Navine also worked on a bioacoustics project looking at the acoustic structure of the ʻōmaʻo whisper song, a song so quiet it’s barely heard from 10 meters away. This work is published here:

• Innovative microphone transmitter reveals differences in acoustic structure between broadcast and whisper songs of Myadestes obscurus (ʻŌmaʻo) (in the journal Ornithology, Jan. 2025)

A career in bioacoustics

After graduating with her master’s degree in 2021, Navine started her current position as an acoustic bioinformatics specialist at Hart’s LOHE Lab and began working with machine learning for bioacoustic analysis. The work is challenging.

“Passive acoustic monitoring is a powerful, noninvasive way to continuously collect data on vocalizing species, but it produces terabytes of data that need to be efficiently transferred, stored, and analyzed, which poses major technological challenges,” says Navine.

LOHE Lab’s research team began working with BirdNET, a deep neural network developed by Cornell Lab of Ornithology. While BirdNET is powerful in many systems, the team quickly learned that the Hawaiian birds are “tricky.”

“They have a lot of acoustic similarities and engage in mimicry quite often, presenting a unique challenge for machine learning classifiers,” explains Navine.

To solve this problem, the team collaborated with Tom Denton of Google Bioacoustic Research Group and one of the developers of Perch, the best birdsong classifier for Hawaiian birds to date.

“Together we developed a method for analyzing bioacoustic data quickly, efficiently, and accurately,” says Navine. “We’ve shown our metric, call density, a measure of the proportion of the soundscape that contains target species vocalizations, is strongly correlated with bird abundance on the landscape.”

Navine presented her research on this topic at the 2024 Hawaiʻi Conservation Conference.

The work is published here:

• All thresholds barred: direct estimation of call density in bioacoustic data (in the journal Frontiers in Bird Science, April 2024)
• Counting the chorus: A bioacoustic indicator of population density (in the journal Ecological Indicators, Dec. 2024)

Moving forward with the help of AI

In 2024, after a decade of Prof. Hart and his research teams of students and technicians trying to improve how researchers track Hawaiʻi forest bird population trends, LOHE Lab is now turning to artificial intelligence to help with monitoring and managing the gigantic database. One of the primary methods Hart uses in his research is recording birdsong through strategically placed recorders within Hawaiian forests. The lab is now using AI to help process its huge database of bird song.

• UH Hilo biologist Patrick Hart’s lab develops AI to analyze forest bird soundscapes (UH Hilo Stories, Oct. 2024)

Given this big change in technology, Navine is currently serving as a bridge between ecology and the computer science needs in Hart’s lab.

“While I don’t have any formal training in programming or computer science, I am self-taught and have a knack for it, at least enough to train a classifier,” she says. “I help to make machine learning classifiers accessible for analyzing datasets that are too large to be manually assessed.”

Using Perch, Navine also conducts bioacoustic analysis of passively collected soundscapes for monitoring bird populations across the Hawaiian Islands. She’s leading several contracted projects for landowners who are interested in understanding how different land management techniques may impact bird species in the area, as well as research projects looking at changes in bird populations over time in response to factors such as disease pressure and climate change. The list of projects is long, ranging from forest birds to seabirds, from Hawaiʻi Island to Nihoa, from mauka to makai (mountain to sea).

Navine also provides the technical skills needed for students and lab volunteers wanting to use machine learning classifiers to analyze their own bioacoustic data using Perch on Jetstream2, a national cloud computing resource supported by the National Science Foundation and available to researchers in the University of Hawaiʻi System. The resource is designed to be easily used by researchers with limited experience in cloud computing; users can understand immense amounts of data from a laptop or tablet.

“Jetstream2 has been absolutely critical to scaling up our bioacoustic analysis capacity,” says Navine. “The data from all of our various projects are analyzed on Jetstream2 virtual machines. We simply could not have accomplished what we have so quickly without a high-performance computing resource, and having access to such a resource for free has been incredibly valuable to the lab.”

Navine says that at any given time the LOHE Lab research team has eight different virtual machines running, each with terabytes of bioacoustic data stored on them.

“I can provide access to these virtual machines to our lab members and alumni via Web browsers, opening up the analysis process to many simultaneously,” she says.

And the results are eye-opening. Using Perch, the LOHE Lab team has been able to detect endangered birds outside of their known ranges, an exciting discovery for the researchers.

“Knowing where endangered birds are is critical to protecting them, and because we can now analyze massive bioacoustic datasets efficiently, we can better determine home ranges with sound,” says Navine.

Giving back to her alma mater and the environment

With her career firmly launched at LOHE Lab, UH Hilo alumna Navine works with most of the UH Hilo students who come through the lab, particularly the graduate students.

Currently, she is working with a grad student who is assessing forest bird abundance in response to landscape scale mosquito (the primary disease vector threatening native birds) control on Maui, and another who is using bioacoustics to find endangered seabirds on Maunakea.

Another example is recent work with a group of undergraduate students to track juvenile bird abundance via bioacoustics at Hakalau Forest National Wildlife Refuge.

Looking to the future, Navine says the biggest goal of LOHE Lab is to expand the passive acoustic monitoring network at Hakalau Forest National Wildlife Refuge, the most intact and disease free native forest on Hawaiʻi Island.

“We want to use recorders as our ears in the forest to serve as an early warning system for possible mosquito incursions into the refuge and subsequent bird population decline,” she says.

Story by Susan Enright, a public information specialist for the Office of the Chancellor and editor of UH Hilo Stories. She received her bachelor of arts in English and certificate in women’s studies from UH Hilo.