Adao Henrique Rosa Domingos
IPBio-Biodiversity Research Institute, Atlantic Forest Research Center/Betary Reserve, São Paulo, Brazil

Background

The Atlantic Forest is one of the most biodiverse and ecologically significant regions in Brazil, but dense vegetation and elusive fauna make it challenging to monitor wildlife through direct observation. To address this, IPBio launched a long-term bioacoustics project using Song Meter SM4 recorders to track species through sound. The goal: to build a comprehensive sound bank of species vocalizations—particularly birds, amphibians, and mammals—and to analyze activity patterns across time and ecosystems.

Objectives

• Train staff and volunteers on bioacoustics technology and software.

• Identify and document vocal species in various forest stages and altitudes.

• Understand daily and seasonal activity patterns of vocalizing species.

• Develop an educational and scientific sound archive.

• Transition from traditional species recognizers to machine-learning classifiers.

Implementation and Methods

Training and Initial Testing:
IPBio translated key sections of the SM4 manual into Portuguese, enabling local biologists and volunteers to use the devices effectively. Biologist Henrique Domingos led early tests across three sites varying in habitat type and altitude, recording birds, amphibians, and mammals.

Monitoring Sites:

• Hill M1 (356m elevation): A remote forest site selected for long-term monitoring due to high biodiversity and presence of Sapajus nigritus (black capuchin monkey).

• Pitfall 5: A site near reserve headquarters with high activity of birds and amphibians; recordings were scheduled every hour for 10 minutes, 24/7.

Data Collection & Equipment:

• SM4 recorders operated during peak activity hours (morning/evening).

• Bushnell camera traps installed alongside to visually confirm animal presence.

• Over 1,000 hours of audio data collected from four distinct locations.

Software & Analysis:
Initially, IPBio used Song Scope to create “recognizers” for species, a manual and time-consuming process. With Wildlife Acoustics’ grant for Kaleidoscope Pro, the lab adopted automated clustering and classifiers—dramatically speeding up analysis and improving accuracy.

Key Results and Outcomes

• Species Inventory: Detected 30+ bird species (including Carpornis melanocephala, a vulnerable species), 2 mammal species (Sapajus nigritus, Alouatta guariba), and 3 amphibians.

• Seasonal and Diurnal Trends: Using 24-hour recordings, IPBio visualized species activity by time of day and season. Birds were most active during daylight; insects dominated at night.

• Education and Outreach: Developed an auditory wildlife archive to educate school groups and reserve visitors. Sounds of elusive animals like Alouatta clamitans (howler monkey) enhanced environmental education experiences.

• Software Efficiency: Kaleidoscope’s clustering technology reduced manual work, allowing interns and researchers to perform year-long analyses within weeks. One intern from Antioch University successfully graphed seasonal frog vocalizations within a month.

Challenges and Adaptive Solutions

• Species Identification Complexity: With hundreds of species vocalizing, confirming IDs was difficult. IPBio collaborated with university researchers across Brazil to assist with identifications.

• Manual Labor with Early Software: Early use of recognizers was labor-intensive; the transition to classifiers and clustering streamlined the workflow.

• Environmental Noise and Overlap: Overlapping vocalizations from multiple species required careful software configuration and validation using camera traps.

Impact and Future Directions

IPBio’s bioacoustic project has become a vital component of biodiversity monitoring in the Atlantic Forest. It supports conservation, science, and education through:

• A growing, curated sound library of forest fauna.

• Data that inform how species respond to environmental changes.

• A replicable monitoring model for other reserves and ecosystems.

Future Plans:

• Continue year-round monitoring at key sites.

• Expand the classifier system for more species.

• Use findings to study impacts of climate change and habitat disturbance.

• Scale the project to include other Brazilian biomes.

Conclusion

IPBio’s integration of bioacoustics, camera trapping, and machine-learning tools has provided an innovative, non-invasive method for monitoring biodiversity in one of the world’s most ecologically important forests. This pioneering work not only contributes to scientific knowledge but also enhances public awareness and supports policy-making for conservation in Brazil.