Using Field Recordings to Study and Document Endemic and Rare Species

Field recordings have become an indispensable tool for biologists, ecologists, and conservationists working to study and document endemic and rare species. These recordings capture the acoustic signatures of animals in their natural habitats, providing a wealth of data that is often impossible to obtain through visual observation. For species that are elusive, cryptic, or active only at night, sound offers a reliable window into their presence, behavior, and population dynamics. As global biodiversity faces unprecedented threats, the ability to monitor and understand these species through non-invasive acoustic methods has never been more critical.

Unlike traditional survey techniques that require direct sightings or trapping, field recordings allow researchers to gather information over large spatial and temporal scales with minimal disturbance. The resulting data can be archived permanently, enabling re-analysis as analytical methods improve and providing a historical baseline for detecting changes in species distributions and vocal behavior. This article explores the importance of field recordings for endemic and rare species, the methods used to collect and analyze them, practical challenges, and the promising future of bioacoustics in conservation.

The Unique Value of Acoustic Data for Endemic and Rare Species

Endemic species are those found only in a specific geographic region, often with small population sizes and specialized habitat requirements. Rare species, whether naturally low in abundance or depleted by human activities, are similarly difficult to study using conventional methods. Acoustic monitoring offers several distinct advantages:

  • Non-invasive detection: Recording devices can be placed in sensitive habitats without causing stress or altering animal behavior, which is especially important for species already under pressure.
  • Continuous surveillance: Automated recording units can operate for weeks or months, capturing vocalizations that might occur only briefly at dawn, dusk, or during specific seasons.
  • Detection of cryptic behavior: Sounds can reveal mating calls, territorial disputes, alarm responses, and other behaviors that are rarely observed visually.
  • Archival potential: Recordings can be revisited years later to apply new analytical techniques or to compare with fresh data, creating a long-term record of acoustic biodiversity.

For many rare and endemic species, the first evidence of their presence in a new location has come from an audio recording rather than a visual sighting. In tropical rainforests, researchers have discovered new populations of frog and bird species entirely through their distinctive calls captured by automated recorders. This has been the case with the Leptodactylus frogs in the Amazon and the elusive Grallaria antpittas in the Andes.

Types of Field Recordings and Equipment

Choosing the right recording approach depends on the target species, habitat, research question, and available budget. The equipment used ranges from simple handheld recorders to sophisticated multi-microphone arrays deployed in remote areas. Proper equipment selection directly affects detection probabilities and data quality.

Portable Recorders

Handheld digital audio recorders, such as those from Zoom, Tascam, or Sony, are widely used for on-the-fly recordings. They are compact, battery-powered, and equipped with built-in microphones or inputs for external microphones. Researchers can use them to make directed recordings of a specific individual or to capture ambient soundscapes. Their portability makes them ideal for surveys in rugged terrain where setting up permanent stations is impractical. However, battery life is typically limited to a few hours, making them better suited for short-duration sampling or targeted recording of known calling events. Many modern portable recorders also support high sample rates (up to 192 kHz) to capture ultrasonic calls of bats or insects, expanding their utility.

Automated Recording Units (ARUs)

ARUs are self-contained devices that can be deployed in the field for extended periods, often programmed to record at specific times (e.g., during peak calling hours) or on a continuous loop. Popular models include the Wildlife Acoustics Song Meter, the Audiomoth (a low-cost open-source option), and the Swift Recorder. These units are ruggedized, weatherproof, and often come with large memory cards and long-lasting battery packs (some can operate for months). ARUs have revolutionized the study of rare species by allowing researchers to cover large areas simultaneously, monitoring multiple sites without needing to be physically present. A single ARU can collect thousands of hours of audio over its deployment, providing a rich dataset for later analysis. The Audiomoth, in particular, costs under $100 and has enabled community-based monitoring projects in resource-limited regions.

External link: Wildlife Acoustics offers a range of ARUs and analysis software widely used in conservation research.

Smartphone Apps and Community Science

With the proliferation of smartphones, mobile apps have become a valuable tool for community scientists and professional researchers alike. Apps like BirdNET, Merlin Bird ID, and iNaturalist’s sound recording feature enable quick identification of species based on their calls. While smartphone recordings are generally lower quality than dedicated recorders, they are accessible and can be used to crowdsource data from many observers. This approach is particularly effective for documenting the presence of rare species across large geographic areas, supplementing formal surveys with observations from hikers, birdwatchers, and naturalists. In some cases, community scientists have reported the first regional records of endangered species through these platforms, leading to targeted research expeditions.

Specialized Array and Directional Recording

For species that are highly mobile or inhabit complex acoustic environments (e.g., dense forests, coral reefs), researchers may use microphone arrays to triangulate sound sources. Directional microphones or parabolic reflectors can isolate calls from a specific individual, reducing background noise. These techniques are especially useful for studying animal behavior and estimating population density through acoustic localization. However, they require more expertise and equipment, and data analysis is computationally intensive. For marine species, hydrophone arrays are used to track vocalizing whales over kilometers, providing real-time data that can mitigate ship strikes.

Data Collection and Field Deployment Strategies

Successful field recording for endemic and rare species requires careful planning of deployment locations, timing, and recording settings. The goal is to maximize the probability of capturing target calls while minimizing power consumption, data storage needs, and environmental interference. Each species has a unique acoustic niche, and mismatches between recording strategy and animal behavior can lead to false absences.

Site Selection and Sampling Design

Researchers typically choose recording sites based on historical occurrence data, habitat suitability models, or local knowledge from indigenous communities. For rare species, a stratified random sampling design can ensure that all habitat types within a potential range are covered. ARUs are often placed along transects or at grid points, with spacing determined by the estimated detection range of the target species’ vocalizations. For example, for a songbird with a call that carries 200 meters, ARUs might be placed every 400 meters to avoid overlap while ensuring coverage. In dense forests where sound attenuates quickly, spacing may need to be reduced to 100 meters. Pilot studies with playback experiments can help calibrate these distances.

Timing and Seasonality

Many endemic species have highly seasonal calling activity, often linked to breeding seasons, rainfall patterns, or lunar cycles. For instance, tropical frogs often call only after heavy rains, while many bird species are most vocal at dawn. Setting ARUs to record during these windows can dramatically increase detection rates while conserving battery and memory. Researchers may use programmable schedules that change throughout the deployment, such as recording for 10 minutes every hour during the day and more frequently at dawn and dusk. For nocturnal species like the Kakapo, a flightless parrot from New Zealand, recorders are set to run continuously during moonlit nights when males perform their booming calls.

Environmental Factors and Sound Attenuation

Acoustic signals can be degraded by wind, rain, running water, and anthropogenic noise such as traffic or machinery. Placing microphones away from direct wind, using windscreens, and choosing sheltered recording locations can improve signal-to-noise ratio. In habitats with high background noise (e.g., waterfalls, urban edges), higher-frequency calls may be obscured, and researchers may need to focus on species with lower-frequency vocalizations that travel farther. Recording during calm, dry conditions is ideal but not always possible in variable climates. When environmental interference is unavoidable, post-processing noise reduction algorithms become essential, though they can introduce artifacts if applied aggressively.

Analyzing Field Recordings: From Raw Audio to Species Identification

Once recordings are collected, the task of extracting meaningful information begins. Raw audio files are often massive (a single ARU deployment can produce hundreds of gigabytes) and require automated or semi-automated analysis. The process typically involves three stages: pre-processing, feature extraction, and classification. The choice of analytical approach depends on the research question, whether it is presence/absence detection, density estimation, or behavioral analysis.

Pre-processing and Noise Reduction

Raw recordings are cleaned to remove or reduce unwanted sounds. High-pass filters can eliminate low-frequency wind rumble, while band-pass filters isolate the frequency range of the target species. Algorithms like spectral subtraction or Wiener filtering can suppress stationary background noise. However, researchers must be cautious not to remove or distort the target signals, especially for rare species with subtle calls. It is common to keep both raw and filtered versions of the data. For marine recordings, specialized filter banks are used to remove noise from snapping shrimp and ship propellers, which can otherwise mask whale calls.

Spectrogram Visualization and Manual Annotation

A spectrogram is a visual representation of sound frequency over time, with intensity indicated by color or brightness. Researchers can scan spectrograms to identify characteristic patterns—distinctive shapes, note durations, and frequency modulations that correspond to particular species. Manual annotation is time-consuming but remains the gold standard for building reference libraries of known calls. Software like Raven Pro, Audacity, and Kaleidoscope (by Wildlife Acoustics) allows users to mark and label vocalizations, creating a training dataset for automated classifiers. For rare species with few recordings, manual annotation is often the only reliable method, and it builds expertise among field researchers.

External link: Raven Pro software from the Cornell Lab of Ornithology is a leading tool for bioacoustic analysis.

Automated Recognition Using Machine Learning

Recent advances in machine learning have revolutionized the analysis of large acoustic datasets. Convolutional neural networks (CNNs) trained on thousands of annotated spectrograms can now identify many species with high accuracy, often exceeding human performance in noisy conditions. Tools like BirdNET, the Google AudioSet, and custom models built with TensorFlow or PyTorch enable researchers to process terabytes of data in hours. For endemic and rare species, these models must be trained with sufficient examples of their calls, which can be a challenge when very few recordings exist. Techniques such as transfer learning and data augmentation help overcome this limitation, allowing models to generalize from related species or synthetically generated calls.

External link: BirdNET is a free AI-powered app and platform for bird sound identification developed by the Cornell Lab of Ornithology.

Acoustic Indices and Soundscape Analysis

Beyond species identification, field recordings can be analyzed using acoustic indices—mathematical measures that describe the complexity, diversity, and intensity of sound over time. Indices such as the Acoustic Complexity Index (ACI), Bioacoustic Index (BI), and Normalized Difference Soundscape Index (NDSI) can serve as proxies for overall biodiversity and habitat health. For rare species, changes in acoustic indices over time can indicate shifts in community composition or the arrival or disappearance of species. This approach is particularly useful when the calls of all species present are not fully known. For example, a sudden drop in ACI in a tropical forest may signal the local extinction of an endemic frog chorus, even if the individual species cannot be identified.

Case Studies: Field Recordings in Action

Real-world examples illustrate the power of field recordings for studying endemic and rare species across different taxa and ecosystems. These cases highlight the diversity of applications, from terrestrial birds to marine mammals.

Hawaiian Honeycreepers

The Hawaiian honeycreepers are a group of endemic birds, many of which are critically endangered or extinct. Researchers from the University of Hawaii used ARUs to survey remote forests on the island of Kauai for the ‘Akikiki and ‘Ake‘e. The recordings revealed the presence of a few remaining individuals in areas that had not been surveyed visually for years. By analyzing call rates and patterns, scientists estimated population densities and identified core territories, guiding habitat protection efforts. The recordings also documented the dawn chorus, providing a baseline for monitoring future changes as avian malaria spreads. In 2023, recordings from these ARUs helped confirm the first successful breeding of ‘Akikiki in a predator-proof enclosure, a milestone for the species' recovery.

Andean Frogs of the Genus Telmatobius

High-altitude frogs in the Andes are among the most threatened amphibians in the world. Many species are endemic to single river systems. A team of herpetologists deployed waterproof ARUs along streams in Peru to capture the underwater and emergent calls of Telmatobius frogs. The recordings revealed previously unknown calling behavior, including vocalizations during the day and in response to changes in water temperature. Acoustic analysis helped differentiate between two morphologically similar species that had overlapping ranges, clarifying their distribution and conservation status. The data also showed that calling activity peaks during the rainy season, allowing conservationists to time tadpole surveys more effectively.

North Atlantic Right Whales

While most examples focus on terrestrial species, marine field recordings are equally important. The North Atlantic right whale is one of the rarest large whales, with fewer than 350 individuals remaining. Autonomous gliders and bottom-mounted hydrophones record their distinctive up-calls over vast areas of the ocean. Real-time analysis of these recordings allows ships to reroute to avoid collisions, a leading cause of mortality. The data also reveals calving grounds and migration corridors that are critical for protective measures. In 2024, an acoustic monitoring network detected a right whale call in an area previously thought abandoned, leading to the discovery of a new calving habitat off the coast of Georgia.

External link: NOAA Fisheries uses acoustic monitoring to protect North Atlantic right whales; read more about their program.

Conservation Applications and Policy Impact

The data from field recordings directly informs conservation decisions for endemic and rare species at multiple levels. Acoustic evidence is increasingly accepted by permitting agencies and international bodies as reliable documentation for species presence and abundance.

Habitat Protection and Restoration

Acoustic surveys can identify critical habitats that may not be obvious from satellite imagery or ground-based visual surveys. For example, the presence of an endemic bird’s song in a forest fragment indicates that the area still supports a viable population, even if the birds are never seen. Conservation organizations use this evidence to prioritize land acquisition, establish protected areas, or design habitat corridors. In some cases, recordings have revealed that a rare species persists in degraded habitats previously dismissed as unsuitable, prompting restoration efforts. The Cerro de la Muerte harlequin frog in Costa Rica was rediscovered through acoustic monitoring in a secondary forest patch that had been slated for development, leading to a successful land purchase.

Population Monitoring and Trend Detection

Repeating acoustic surveys over years or decades provides robust data on population trends. For species that are rare or have low detectability, traditional field surveys often produce too few observations to estimate trends with confidence. Acoustic data, on the other hand, can yield hundreds of detections per deployment. Researchers use occupancy modeling and distance sampling adapted for acoustic data to estimate abundance and detection probability. This allows early detection of population declines before they become catastrophic. For the endangered Hula painted frog in Israel, long-term acoustic monitoring showed a 40% decline in calling activity over five years, triggering emergency conservation measures.

Illegal Activity Detection

Field recordings can also pick up sounds of illegal activities such as gunshots (poaching), chainsaws (illegal logging), or motor vehicles entering restricted areas. Some conservation teams use recordings not only for species identification but also as a security tool. AI systems can be trained to recognize gunfire or chainsaw noise, triggering alerts for rapid response teams. For rare species targeted by poachers, this integrated approach provides simultaneous surveillance of the species and the threats it faces. In the Amazon, acoustic sensors have been used to detect illegal gold mining activity near endemic primate populations, allowing rangers to intervene before habitats are destroyed.

Challenges and Limitations

Despite its many advantages, acoustic monitoring for endemic and rare species is not without difficulties. Practitioners must navigate technical, logistical, and analytical hurdles that can affect data quality and interpretation.

Data Volume and Storage

Continuous recording generates enormous amounts of data. A single ARU recording in stereo at 48 kHz can produce over 20 GB per week. For a project with dozens of units, this quickly becomes unmanageable without cloud storage or on-device processing. Many modern ARUs now offer onboard compression or event-triggered recording (only recording when sound levels exceed a threshold), but these features can miss quiet calls of rare species. Researchers must balance data quality with storage capacity and budget. Cloud-based solutions like those offered by EarthRanger or Microsoft's Planetary Computer can help, but they require reliable internet in remote areas.

Training Data Scarcity

Machine learning models require many labeled examples to perform well. For many endemic and rare species, few or no high-quality recordings exist. This "cold start" problem limits the application of automated recognition. Approaches such as active learning (where the model asks for human verification on uncertain detections) and using calls from closely related species as surrogates can help, but they are imperfect. Collaborative efforts to build shared reference libraries, like the Macaulay Library at Cornell and the Xeno-canto database, are essential for filling these gaps. For some species, researchers have resorted to playing back recorded calls to elicit responses and record them with higher quality.

External link: Xeno-canto is a community database of bird sounds from around the world, including many endemic species.

Background Noise and Signal Overlap

In dense acoustic environments, calls from multiple species can overlap, especially during the dawn chorus. Rare species with quiet or infrequent calls are easily masked. Physical separation of microphones, such as using unattended overlapping arrays to triangulate, can help isolate signals, but adds complexity. Advanced unmixing algorithms based on blind source separation are an active area of research, but they require high signal-to-noise ratios that are not always achievable in the field. For marine environments, ship noise is a growing problem that can mask whale calls over kilometers; researchers are developing real-time adaptive filtering to mitigate this.

Equipment Reliability and Theft

Deploying expensive recording equipment in remote or public areas carries risk of theft, damage by animals, or failure due to extreme weather. Researchers often use camouflage, steel cables, and lockable enclosures. Redundant recorders at key sites can mitigate data loss, but this increases costs. Battery life remains a limitation, especially in cold climates where lithium-ion performance drops. Solar-powered options are emerging but add weight and complexity. In tropical forests, ants and rodents have been known to chew cables or fill enclosures with debris, so protective measures must account for local fauna.

Future Directions and Emerging Technologies

The field of bioacoustics is evolving rapidly, offering new capabilities for studying endemic and rare species. Innovations in hardware, software, and community engagement are lowering barriers and expanding possibilities.

Real-Time Acoustic Monitoring with Edge Computing

Advances in microprocessors and energy-efficient AI chips allow on-device processing. Small, low-power recorders can now run machine learning models in the field, identifying species in real time and transmitting results via satellite or cellular networks. This reduces the burden of data storage and enables immediate conservation actions, such as closing a fishery when an endangered whale call is detected. Projects like the Earth Ranger program are testing these devices in national parks. The ability to receive alerts within minutes of a rare species detection can transform rapid response conservation.

Multimodal Sensor Integration

Combining audio with other sensors (camera traps, temperature loggers, humidity sensors, light meters) provides a richer picture of the environment. For example, a rare frog’s calls recorded near a stream can be correlated with water temperature and flow rate to understand breeding triggers. Integration of acoustic data with satellite imagery helps link species presence to landscape-level changes like deforestation or fire scars. In the cloud forests of Ecuador, integrated sensor networks have shown how microclimate shifts affect the calling behavior of endemic glass frogs, informing climate adaptation strategies.

Passive Acoustic Monitoring for Invertebrates and Plants

While most bioacoustics focus on vertebrates, new research shows that insects, spiders, and even plants produce sounds that can be used for identification. For rare endemic insects like the Lord Howe Island stick insect, recordings of stridulation can help locate remaining populations. Underwater microphones (hydrophones) are increasingly used to monitor rare freshwater invertebrates and fish. This expansion of acoustic monitoring into less-studied taxa will likely reveal new species and improve ecosystem understanding. The sounds of plants, such as cavitation in xylem vessels during drought, may also serve as early warning signals for habitat stress.

Citizen Science and Global Collaboration

Platforms like iNaturalist, eBird, and Xeno-canto now support uploading of sound recordings alongside photos. Citizen scientists in biodiversity hotspots can contribute recordings of rare species, providing data from areas that researchers cannot visit. The rise of low-cost ARUs like the Audiomoth (under $100) has democratized access to professional-grade recording. A global network of volunteers can help monitor endemic species over large geographic areas, enabling early detection of population declines or range shifts due to climate change. In a landmark project, thousands of volunteer recordings from the Amazon helped map the distribution of the endemic white-bellied spider monkey, previously thought to be confined to a small area.

Conclusion

Field recordings have moved from a niche technique to a mainstream tool in the study and conservation of endemic and rare species. By capturing the acoustic life of habitats, researchers can detect species that would otherwise remain invisible, monitor populations over time, and gather behavioral data with minimal disturbance. The combination of affordable automated recorders, powerful machine learning analysis, and global citizen science networks is opening new frontiers in bioacoustics. As technological barriers continue to fall, the role of sound in documenting and protecting the world’s most vulnerable species will only grow. For conservationists, every recorded call is not just a data point—it is a sign of resilience and a call to action.