What Is Bioacoustics? Principles, Passive Acoustic Monitoring & Marine Applications

Bioacoustics is the interdisciplinary branch of science that investigates sound production, dispersion, and reception in biological organisms. Combining principles from acoustic physics, animal behavior, neurobiology, and ecology, bioacoustics examines how non-human animals use sound to navigate, forage, maintain social cohesion, and defend territory.

While terrestrial bioacoustics explores avian songs, primate calls, and insect stridulation, marine bioacoustics occupies a particularly vital role in planetary science. Because seawater absorbs and scatters optical wavelengths within tens of meters, sound is the primary sensory modality through which marine mammals perceive their environment across kilometers of ocean.

Physical Fundamentals of Underwater Acoustics

Understanding bioacoustics requires examining how physical oceanographic properties shape sound transmission through the water column:

1. Speed of Sound and Environmental Variables

The speed of sound in seawater (c, in m/s) is governed dynamically by temperature (T, in °C), salinity (S, in practical salinity units / PSU), and hydrostatic depth (z, in meters). A widely used empirical formula derived by Medwin (1975) for typical oceanic ranges (0 ≤ T ≤ 35°C, 0 ≤ S ≤ 45 PSU, z ≤ 1,000 m) is:

c ≈ 1449.2 + 4.6T - 0.055T² + 0.00029T³ + (1.34 - 0.01T)(S - 35) + 0.016z

• Surface Layer: Solar heating creates a warm, well-mixed surface layer where temperature dominates, resulting in higher sound speeds.
• Thermocline: Beneath the mixed layer, water temperature drops steeply with depth, causing sound speed to decline to a minimum value.
• Deep Ocean: Below the thermocline (typically 800–1,200 m depending on latitude), temperature stabilizes at 2–4°C, and steadily rising hydrostatic pressure (+0.016 m/s per meter depth) drives sound speed back upward.

This depth-dependent velocity gradient creates the SOFAR (Sound Fixing and Ranging) channel. Acoustic energy refracted toward the sound-speed minimum is channeled by continuous upward and downward refraction, avoiding lossy sea-surface and benthic reflections. While low-frequency signals (such as baleen whale calls below 100 Hz) can propagate across hundreds to thousands of kilometers within this channel, actual propagation range is strongly constrained by channel axis depth, bathymetric features (seamounts and continental shelves), and oceanographic boundary conditions.

2. Frequency-Dependent Absorption and Chemical Relaxation

As acoustic waves travel through seawater, acoustic energy is dissipated as heat via viscous absorption and chemical relaxation processes:

• Boric Acid Relaxation (B(OH)₃): Dominates chemical sound absorption at frequencies below ~1 kHz.
• Magnesium Sulfate Relaxation (MgSO₄): Dominates sound absorption across the intermediate frequency band (~1 kHz to ~100 kHz).
• Pure Water Viscosity: Becomes the primary absorption mechanism at frequencies above ~100 kHz.

Consequently, attenuation scales exponentially with frequency:

• Ultrasonic Signals (above 40 kHz): Dolphin echolocation clicks attenuate rapidly over hundreds of meters to a few kilometers, providing high spatial resolution for biosonar while constraining detection range.
• Mid-Frequency Signals (4–24 kHz): Delphinid whistles propagate effectively over 1 to 5 kilometers in typical shallow and pelagic environments.
• Low-Frequency Signals (below 100 Hz): Fin and blue whale infrasonic calls experience low attenuation, supporting long-range communication when channeled by favorable sound-speed profiles.

Acoustic Wave Propagation in Ocean Media
───────────────────────────────────────────────────────────────────
Infrasound (<20 Hz)        Tonal Whistles (4–24 kHz)   Echolocation (>40 kHz)
[ Baleen Whale Moans ]     [ Dolphin Whistles ]        [ Biosonar Clicks ]
   ◄── Basin-scale ──►       ◄── 1 to 5 km range ──►     ◄── 50 to 500 m ──►
   Low attenuation           Moderate attenuation        Rapid absorption

Passive Acoustic Monitoring (PAM)

Because marine mammals spend the overwhelming majority of their lives submerged and out of sight, visual surveys miss significant portions of animal activity. Passive Acoustic Monitoring (PAM) uses autonomous hydrophones to listen to marine soundscapes non-invasively, day and night, across seasonal and multi-year time scales.

Hydrophone Instrumentation and Sensitivity

Hydrophones employ piezoelectric ceramic sensors that generate a voltage proportional to acoustic pressure fluctuations. Calibrated hydrophone sensitivity is typically expressed in decibels relative to one volt per micropascal (dB re 1 V/μPa). Modern PAM deployments utilize several configurations:

1. Bottom-Moored Autonomous Recorders: Moored to the sea floor with acoustic releases, recording continuously or on duty cycles for months or years. For delphinid research, recorders require high ultrasonic sampling rates (e.g., 192 kHz to 500 kHz) to record clicks with frequencies above 100 kHz without aliasing.
2. Towed Hydrophone Arrays: Streamed behind research vessels to localize vocalizing cetaceans in real time using time-difference-of-arrival (TDOA) beamforming.
3. Autonomous Marine Gliders & Floats: Low-noise buoyant platforms that profile ocean waters while recording acoustic data.
4. Drifting Vertical Arrays: Multi-channel hydrophone arrays deployed near focal groups to isolate calls from individual animals.

Historical archives such as the Watkins Marine Mammal Sound Database (curated by the Woods Hole Oceanographic Institution) and modern benchmarks like the DCLDE Marine Mammal Dataset preserve decades of these standardized PAM recordings.

Marine Mammal Acoustic Repertoires

Odontocetes (toothed whales, dolphins, and porpoises) produce diverse acoustic signals categorized into three functional classes:

                  Delphinid Acoustic Repertoires
                                │
       ┌────────────────────────┼────────────────────────┐
       ▼                        ▼                        ▼
[ Frequency-Modulated ]   [ Broadband Ultrasonic ]   [ Burst-Pulsed ]
  Tonal Whistles            Echolocation Clicks        Packets
  • Fundamental: 4–24 kHz   • Center: 40–130+ kHz      • Inter-Click: <5 ms
  • Social identity         • Biosonar navigation      • Agonistic & social
  • Long-term stability     • Foraging targets         • Complex harmonics

1. Frequency-Modulated Tonal Whistles

Narrowband whistles with fundamental frequencies typically between 4 kHz and 24 kHz. In common bottlenose dolphins (Tursiops truncatus), longitudinal studies by researchers including Dr. Laela S. Sayigh and Dr. Peter L. Tyack (

) established that individuals develop unique signature whistles that broadcast individual identity and remain stable across decades.

2. Broadband Echolocation Clicks

Short-duration (tens of microseconds) high-intensity impulses with peak frequencies extending from 40 kHz to above 130 kHz. Dolphins use clicks for biosonar navigation, obstacle avoidance, and prey detection, adjusting inter-click intervals as they close in on targets.

3. Burst-Pulsed Vocalizations

High-rate trains of clicks separated by intervals so short (less than 5 ms) that the human ear perceives them as a single tonal sound (buzzes, squawks, and barks). Frequently produced during intense social, aggressive, or alarm encounters.

In larger odontocetes, such as the sperm whale (Physeter macrocephalus), acoustic repertoires are dominated by rhythmic click patterns called codas. Recent research by the Project CETI consortium (

) has uncovered combinatorial variations and rubato tempo dynamics across sperm whale clans.

The Computational Revolution in Bioacoustics

Classical bioacoustic workflows required researchers to manually inspect sonograms—a process that becomes intractable when multi-year PAM deployments collect hundreds of terabytes of audio.

Modern computational bioacoustics leverages machine learning across three primary paradigms:

1. Convolutional Classifiers: Systems like BirdNET convert audio into Short-Time Fourier Transform spectrograms and train deep convolutional neural networks to classify vocalizations across thousands of wildlife species.
2. Self-Supervised Bioacoustic Embeddings: Foundation models (such as Google Perch and AVES) extract general acoustic feature vectors directly from soundscapes without manual segmentation.
3. Discrete Neural Codecs & Sequence Transformers: Exploratory models like DolphinGemma employ residual vector quantization (RVQ) to discretize high-frequency hydrophone audio into acoustic tokens, allowing sequence transformers to discover statistical motifs across longitudinal field archives.

Bioacoustics bridges physical acoustics and ecological science, enabling researchers to monitor biodiversity, safeguard endangered marine habitats from anthropogenic noise, and investigate the evolution of communication across the animal kingdom.