Echolocation in Sea Animals: How Marine Mammals Navigate the Deep

Echolocation is an active biological sonar system used by odontocetes, the toothed whales and dolphins, to navigate, hunt, and interpret their underwater environment by emitting high-frequency sound pulses and analyzing the returning echoes. This article explains how echolocation works across marine mammal species, compares the echolocation abilities of dolphins, whales, and other marine mammals, and describes how researchers study these signals using passive acoustic monitoring and machine learning. The content is written for students, researchers, life-science professionals, and informed general readers who want a practical understanding of marine mammal biosonar, including its evolutionary context, measurement methods, and conservation applications.

What Is Echolocation and Which Marine Mammals Use It

Echolocation is the process by which an animal emits sound and uses the reflected echo to form an acoustic image of its surroundings. Among marine mammals, echolocation is a defining feature of odontocete cetaceans, the group that includes dolphins, porpoises, and toothed whales such as sperm whales and beaked whales. The evolutionary success of this lineage is closely tied to the development of echolocation, which uses high-frequency sound produced in the forehead and detected by the cochlea [9]. The melon, a rounded lipid structure located in the forehead between the blowhole and the tip of the head, plays a primary role in focusing and transmitting these echolocation signals [14].

Not all marine mammals echolocate. Pinnipeds, the group that includes seals, sea lions, and walruses, do not possess an advanced active biosonar system. The evidence for pinniped echolocation remains unconvincing, and the obligate amphibious functioning of the pinniped auditory system appears to constrain the evolution of the highly acute, aquatic, high-frequency sound production and reception systems required for underwater echolocation [7]. Instead, pinnipeds have evolved enhanced visual, tactile, and passive listening skills to forage, navigate, and avoid predators underwater [7].

The distinction matters for researchers and conservation professionals. When monitoring marine mammals acoustically, the presence or absence of echolocation clicks can indicate which species are present and what they are doing. For example, passive acoustic monitoring datasets that record vocalizations and echolocation signals are used to track species distributions and support conservation planning [6].

How Odontocetes Produce and Receive Echolocation Signals

Odontocetes generate echolocation clicks through a complex system of air sacs and phonic lips in the nasal passage, located below the blowhole. The sound is focused and projected forward through the melon, which acts as an acoustic lens. The returning echoes are received primarily through the lower jaw, which conducts sound to the inner ear. The cochlea, the auditory portion of the inner ear, is the final detection site for these high-frequency signals [9].

The melon serves an acoustic function and participates in immune responses. Research on the striped dolphin (Stenella coeruleoalba) has shown that the melon contains immune cells, including macrophages, and that melon adipocytes themselves react to antibodies used to characterize immune function [14]. This finding indicates that the melon may participate in immune responses, a consideration for researchers studying the health of wild and captive dolphins. Contaminants such as heavy metals can accumulate in the fatty tissues of cetaceans, including the melon and blubber, and these accumulations can alter immune function and stimulate inflammatory responses [14].

The cochlear morphology of odontocetes reflects the acoustic demands of their environments. A comparative study of toothed whale cochleae identified convergent evolution in cochlear shape among species that occupy similar acoustic habitats [9]. Sperm whales and beaked whales, which forage in the deep ocean, show convergent cochlear morphology, suggesting that the extreme acoustic environment of the deep ocean constrains cochlear shape [9]. Habitat type and dive type were significantly correlated with membership in this convergent group [9]. This finding supports the use of cochlear morphology to predict the ecology of extinct cetaceans [9].

At a Glance: Echolocation Capabilities Across Marine Mammal Groups

The table below summarizes the echolocation status and key characteristics of major marine mammal groups. This comparison is useful for students and professionals who need a quick reference for species identification and acoustic monitoring decisions.

The Physics of Echolocation Clicks and Echo Reception

Echolocation clicks are short-duration, broadband signals. The frequency content, duration, and inter-click interval vary by species and by behavioral context. When a click encounters an object, part of the sound energy reflects back to the animal as an echo. The time delay between click emission and echo reception provides information about distance, while the spectral characteristics of the echo provide information about the size, shape, material, and internal structure of the object.

The acoustic properties of the target affect the echo. A simulation study based on cross-modal matching experiments with a bottlenose dolphin (Tursiops aduncus) examined how dolphins discriminate objects that differ only in material composition [12]. The study used four types of objects: water-filled PVC pipes, air-filled PVC pipes, foam ball arrays, and PVC pipes wrapped in closed-cell foam. The dolphin's matching accuracy differed significantly across these cases, and finite element simulations of the click interaction with the objects provided possible explanations for the performance differences [12]. This research demonstrates that material composition, beyond shape and size, influences the echo features available to the echolocating dolphin.

The frequency-dependent absorption of sound in seawater is a critical factor in echolocation range. Higher frequencies attenuate more rapidly than lower frequencies, so the effective range of an echolocation click depends on its spectral content. Researchers modeling the propagation of marine mammal echolocation clicks for population density estimates have examined how simplifying assumptions about click spectra affect the estimated effective detection area (EDA) [3]. Common assumptions include approximating click spectra as flat energy spectra and neglecting frequency-dependent sound absorption within the click bandwidth [3]. These assumptions can bias density estimates. For Blainville's beaked whales, the estimated EDA differed by up to a factor of 2 depending on the spectral energy distribution of the clicks [3]. Relative density bias due to narrowband assumptions ranged from 5% to more than 100%, depending on species, detector settings, and noise conditions [3].

Comparing Echolocation Across Dolphin and Whale Species

Different odontocete species produce echolocation clicks with distinct characteristics. These differences allow researchers to classify species from acoustic recordings, but the variability within and between species presents challenges for automated classification.

A large-scale study of dolphin echolocation clicks in the Gulf of Mexico applied an automated network-based classification method to 52 million clicks detected across five monitoring sites over two years [13]. The method incorporated multiple click characteristics, including spectral shape and inter-click interval distributions, to distinguish within-type from between-type variation [13]. Seven distinct click types were identified, one associated with Risso's dolphin and six not yet identified [13]. All types occurred at multiple monitoring locations, but their relative occurrence varied between continental shelf and slope locations [13]. Comparisons with clicks detected on towed hydrophone arrays in the presence of visually identified delphinid species suggested potential species identities for some click types [13].

Deep learning approaches have also been applied to dolphin echolocation click classification. A study comparing Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) for classifying dolphin echolocation clicks into two species groups in a high-background-noise environment found that both models performed well [11]. The highest performance was achieved by a CNN fed with spectrograms (F1 score 97%) and an RNN fed with raw data (F1 score 96%) fitted with Gated Recurrent Units [11]. The RNN showed excellent performance with reduced processing time and storage, and deep learning automatically extracts effective features from raw waveforms without separate pre-processing steps [11].

Sperm whale and dolphin echolocation clicks serve as bioindicators for monitoring marine ecosystems [17]. Detection of click signals provides information on species abundance, behavior, and responses to environmental changes [17]. A survey of detection techniques for sperm whale and dolphin echolocation clicks, covering methods from 2002 to 2023, categorized approaches into feature analysis (phase, inter-click interval, duration), frequency content, energy-based detection, supervised and unsupervised machine learning, template matching, and adaptive detection [17]. The survey also identified open-access platforms for click detection and databases available for testing [17].

Echolocation in Deep-Diving Whales: Sperm Whales and Beaked Whales

Sperm whales (Physeter macrocephalus) and beaked whales are deep-diving odontocetes that rely on echolocation to forage in the mesopelagic and bathypelagic zones, where light does not penetrate. Their echolocation clicks are adapted to the acoustic conditions of the deep ocean.

The cochlear morphology of sperm whales and beaked whales has converged, likely due to the constraints of the deep ocean acoustic environment [9]. This convergence is significant enough that cochlear morphology can be used to predict the ecology of extinct cetaceans [9]. The convergent regimes identified include True's beaked whale (Mesoplodon mirus) and Cuvier's beaked whale (Ziphius cavirostris), sperm whales and all other beaked whales sampled, and pygmy and dwarf sperm whales (Kogia breviceps and Kogia sima) with Dall's porpoise (Phocoenoides dalli) [9].

Estimating population density of beaked whales from passive acoustic monitoring requires modeling the effective detection area around each hydrophone [3]. The modeling approach uses the passive sonar equation and applies detectors to simulated clicks injected into measurements of background noise [3]. The choice of spectral assumptions can substantially bias density estimates, with relative bias ranging from 5% to more than 100% depending on species, detector settings, and noise conditions [3]. Researchers and monitoring programs must account for these biases when interpreting click-based density estimates.

Echolocation and Foraging Behavior in Dolphins

Echolocation is not a fixed behavior. Dolphins adjust their echolocation use based on the foraging context and the behavior of other animals. A study of cooperative foraging between artisanal fishers and wild dolphins in southern Brazil revealed that dolphins modify their active foraging echolocation to match the time it takes for nets to sink and close over mullets, but only when fishers respond to the dolphins' foraging cues appropriately [8]. When dolphins approach the fishers' nets closely and cue fishers in, they dive for longer and adjust their echolocation timing [8]. This foraging synchrony benefits both predators, with cooperative dolphins gaining approximately 13% survival benefits by minimizing spatial overlap with bycatch-prone fisheries [8].

The study also documented that recent declines in mullet availability are threatening these benefits by reducing the foraging success of net-casting fishers and increasing the exposure of dolphins to bycatch in alternative fisheries [8]. A numerical model parameterized with empirical data predicted that environmental and behavioral changes are pushing this human-dolphin cooperation toward extinction [8]. Two conservation actions targeting fisher behavior were proposed to prevent the erosion of this century-old fishery [8].

Killer whales (Orcinus orca) also show differences in sonar use between ecotypes. Fish-eating and mammal-eating killer whales differ in their use of echolocation, reflecting the different acoustic challenges of hunting fish versus marine mammals [20]. Mammal-eating killer whales face the challenge that their prey may detect the echolocation signals, making echolocation a mixed blessing [20]. This behavioral difference is relevant for acoustic monitoring and for understanding the ecological niches of killer whale populations.

Passive Acoustic Monitoring: How Researchers Study Echolocation

Passive acoustic monitoring (PAM) is a common approach to monitor marine mammal populations that use sound for navigation, feeding, and communication [11]. PAM involves deploying hydrophones, either on autonomous seafloor recorders, towed arrays, or widely dispersed monitoring stations, to record the sounds produced by marine mammals. The resulting datasets are often large and benefit from automated detection and classification algorithms [11][4].

The effective detection area (EDA) is a key concept in PAM. It is the area around each receiver over which vocalizations are detected [3]. Estimating the EDA is necessary to convert click detections into population density estimates [3]. In the absence of auxiliary measurements, the EDA can be modeled using the passive sonar equation [3]. However, common simplifying assumptions, such as approximating click spectra as flat and neglecting frequency-dependent absorption, can bias density estimates [3].

Performance metrics are essential for evaluating automated detection and classification algorithms. Four metrics are commonly used: receiver-operating-characteristic (ROC) curves, detection-error-trade-off (DET) curves, precision-recall (PR) curves, and cost curves [4]. These metrics were applied to the generalized power law detector for blue whale D calls and the click-clustering neural-net algorithm for Cuvier's beaked whale echolocation click detection [4]. Detection class imbalance, particularly the situation of rare occurrence, is common in long-term PAM datasets and affects the performance of ROC and DET curves [4]. PR curves overcome this shortcoming when calculated for individual detections and do not rely on the reporting of true negatives [4]. Cost curves provide additional insight on the effective operating range for the detector based on the a priori probability of occurrence [4]. Using more than a single metric is helpful in understanding the performance of a detection algorithm [4].

Machine Learning for Echolocation Click Classification

Machine learning has become a standard tool for classifying echolocation clicks in PAM datasets. The approaches range from unsupervised network-based clustering to supervised deep learning with neural networks.

The automated classification method developed for Gulf of Mexico dolphin clicks used an unsupervised network-based approach that simulates how a human analyst categorizes click types [13]. Clusters of similar clicks were identified by incorporating multiple click characteristics, including spectral shape and inter-click interval distributions, to distinguish within-type from between-type variation [13]. Once click types were established, an algorithm classified novel detections using existing clusters [13]. This method proved effective for rapid, unsupervised classification of large datasets [13].

Deep learning approaches have shown high performance in challenging acoustic environments. A comparison of CNNs and RNNs for dolphin echolocation click classification in a high-background-noise environment found that a CNN fed with spectrograms achieved an F1 score of 97% and an RNN fed with raw data achieved an F1 score of 96% [11]. The RNN's ability to work with raw data reduced processing time and storage requirements [11]. The study recommended using such models in marine environments where background noise levels exhibit high spatial and temporal variance [11].

A comprehensive whistle classification framework for Indo-Pacific bottlenose dolphins (Tursiops aduncus) tested five CNN architectures and reported performance using mean Average Precision (mAP) [15]. The study introduced noise at defined signal-to-noise ratio levels to assess classifier stability and used Bellhop channel simulation to construct channel impulse responses for data augmentation [15]. The simulated data enhanced the robustness of the classification model [15]. This work provides tools for marine biologists and researchers specializing in animal acoustics and contributes to conservation and management efforts for dolphin populations [15].

For beaked whale echolocation pulses, an open-source machine learning classification approach called BANTER has been developed [16]. The method is available for researchers who need to classify beaked whale echolocation pulses in PAM datasets [16].

Bio-Inspired Sonar: Learning from Marine Mammal Echolocation

Marine mammal echolocation has inspired engineering applications in sonar design. The covertness of active sonar is an important issue, and researchers have studied sonar signal waveform design to improve covertness [5]. Many marine mammals produce call pulses for communication and echolocation, and existing interception systems typically classify these biological signals as ocean noise and filter them out [5]. Based on this observation, a bio-inspired covert active sonar strategy was proposed that uses true sperm whale call pulses as sonar waveforms to ensure camouflage [5].

The bio-inspired approach designed a range and velocity measurement combination (RVMC) using two true sperm whale call pulses that had excellent range resolution and large Doppler tolerance [5]. Range and velocity estimation methods were developed based on the RVMC, and correlation technology was used in the sonar receiver to confirm the start and end time of sonar signals and their echoes [5]. The RVMC was embedded into a true sperm whale call-train to improve the camouflage ability of the sonar signal-train [5]. Experimental results verified the performance of the proposed method [5].

A beluga whale (Delphinapterus leucas) inspired echolocation model was developed by observing beluga sound recordings [18]. The study used a signal extracting method with a correction factor for inter-click interval to acquire the parameters of click trains [18]. The extracted clicks were analyzed in the time and frequency domain, and a joint pulse-frequency representation was undertaken to provide a 2D energy distribution for an echolocation click train [18]. The results indicated that a click train can adjust its energy distribution by using a multi-component signal structure [18]. A finite element model was built to reproduce target discrimination by the bio-inspired click train, and numerical results indicated that the bio-inspired click train could enhance the echo-response by concentrating energy into frequency bins for extracting target features effectively [18]. This proof-of-concept study suggests that click train models could be dynamically controlled to match target properties, showing a promising way to use various types of echolocation click trains to interrogate different features of targets with man-made sonar [18].

On-Board Acoustic Recording: Studying Echolocation from the Animal's Perspective

Technological advances have enabled researchers to develop miniature devices, or tags, that record an animal's behavior directly on the animal itself [10]. On-board audio recordings have become more common following pioneering work in marine mammal research [10]. These recordings allow researchers to study when animals are calling, how they adjust their behavior, and what acoustic parameters they change and how [10].

On-board acoustic recordings also enable studies of foraging behavior, social interactions, and environmental acoustics [10]. While the review of on-board acoustic recording techniques focused primarily on bats, the general ideas and concepts are applicable to many animals, including marine mammals [10]. The active-sensing, echolocating lifestyle of bats allows many approaches to a multi-faceted acoustic assessment of their behavior, and these approaches can be adapted for marine mammal research [10].

For marine mammals, on-board acoustic recording can reveal how echolocation use changes during different phases of a foraging dive, how dolphins coordinate their echolocation with other group members, and how environmental noise affects echolocation behavior. This information is valuable for understanding the ecological needs of marine mammals and for designing conservation measures.

Practical Steps for Studying Marine Mammal Echolocation

Researchers and monitoring programs can follow a structured workflow to study marine mammal echolocation. The steps below provide a practical framework for designing and implementing an acoustic monitoring study.

Step 1: Define the Monitoring Objective

Determine whether the goal is species presence detection, population density estimation, behavioral classification, or habitat use mapping. The objective determines the choice of recording equipment, deployment design, and analysis methods. For population density estimation, the effective detection area must be modeled or measured [3]. For species classification, the analysis must account for within-species and between-species variability in click characteristics [13].

Step 2: Select Recording Equipment and Deployment Design

Choose between autonomous seafloor recorders, towed hydrophone arrays, or on-board tags based on the monitoring objective and the target species. Widely dispersed hydrophones have been suggested as a cost-effective method to monitor population densities of echolocating marine mammals [3]. The deployment design must account for the detection range of the target species' clicks, which depends on source level, frequency content, and ambient noise.

Step 3: Collect and Manage Acoustic Data

PAM produces large datasets that benefit from automated analysis [11]. Data management should include standardized file formats, metadata recording, and quality control procedures. The MAMMALS IN PORTUGAL data set provides an example of a publicly available data set that includes vocalization and echolocation records among 13 types of records, with spatial uncertainty of most records ranging between 0 and 100 m [6].

Step 4: Detect and Classify Echolocation Clicks

Apply detection algorithms to identify echolocation clicks in the recordings. The choice of detection algorithm depends on the target species and the acoustic environment. For species with well-characterized clicks, template matching or energy-based detection may be sufficient [17]. For complex environments with high background noise, machine learning approaches such as CNNs and RNNs may be necessary [11]. Evaluate detection and classification performance using multiple metrics, including ROC curves, DET curves, PR curves, and cost curves [4].

Step 5: Estimate Density or Abundance

If the objective is population density estimation, model the effective detection area using the passive sonar equation or simulation-based approaches [3]. Account for the potential bias introduced by spectral assumptions. For Blainville's beaked whales, the estimated EDA differed by up to a factor of 2 depending on spectral assumptions, and relative density bias ranged from 5% to more than 100% [3].

Step 6: Interpret Results in Ecological Context

Interpret echolocation data in the context of species ecology, behavior, and environmental conditions. Echolocation use varies with foraging context, as demonstrated by the differences in sonar use between fish-eating and mammal-eating killer whales [20] and the adjustments dolphins make during cooperative foraging with fishers [8]. Consider the potential for echolocation signals to be detected by prey, which can affect foraging success [20].

Records and Measurements for Echolocation Studies

Accurate record keeping is essential for echolocation research and monitoring. The following measurements and records should be maintained for each monitoring session or study period.

Common Failure Patterns in Echolocation Research and Monitoring

Several common failure patterns can compromise echolocation studies. Recognizing these patterns helps researchers design robust studies and interpret results correctly.

Spectral Assumption Bias

Assuming flat click spectra or neglecting frequency-dependent absorption within the click bandwidth can bias effective detection area estimates and resulting density estimates [3]. The bias can exceed 100% depending on species, detector settings, and noise conditions [3]. Researchers should test the sensitivity of their estimates to spectral assumptions.

Detection Class Imbalance

Long-term PAM datasets often have rare occurrence of target signals, which affects the performance of ROC and DET curves [4]. PR curves overcome this shortcoming when calculated for individual detections [4]. Researchers should use multiple performance metrics to understand algorithm performance [4].

Misclassification of Click Types

Echolocation clicks vary within and between species, and automated classification can misassign clicks to incorrect types [13]. The unsupervised network-based classification method addresses this by incorporating multiple click characteristics to distinguish within-type from between-type variation [13]. Researchers should validate classification results with independent data, such as visual species identification from towed hydrophone arrays [13].

Ignoring Behavioral Context

Echolocation use varies with behavioral context. Dolphins adjust their echolocation during cooperative foraging [8], and killer whale ecotypes differ in sonar use [20]. Ignoring behavioral context can lead to incorrect interpretations of acoustic data.

Overlooking Non-Echolocating Species

Not all marine mammals echolocate. Pinnipeds rely on enhanced visual, tactile, and passive listening skills instead of active biosonar [7]. Acoustic monitoring programs that focus exclusively on echolocation clicks will miss pinniped presence and behavior.

Limitations of Echolocation Research

Echolocation research has inherent limitations that researchers and professionals should acknowledge.

Acoustic Environment Constraints

The effective range of echolocation depends on frequency-dependent sound absorption, ambient noise, and target characteristics. Higher frequencies attenuate more rapidly, limiting the range of high-frequency echolocation clicks. The frequency-dependent absorption within the click bandwidth is often neglected in modeling, which can bias results [3].

Species Identification Challenges

Many echolocation click types cannot be assigned to specific species without independent verification. In the Gulf of Mexico study, six of seven click types were not identified to species [13]. Comparisons with visually identified species from towed hydrophone arrays can suggest potential identities, but verification requires additional data [13].

Behavioral Variability

Echolocation behavior varies with context, including foraging mode, prey type, and social interactions. The mixed blessing of echolocation for mammal-eating killer whales illustrates how prey can detect echolocation signals, potentially affecting foraging success [20]. Researchers must account for behavioral variability when interpreting acoustic data.

Evolutionary Constraints

Echolocation has evolved within specific ecological and morphological constraints. The obligate amphibious functioning of the pinniped auditory system appears to have prevented the evolution of advanced echolocation in pinnipeds [7]. Understanding these constraints helps researchers predict which species are likely to echolocate and how their echolocation systems are adapted to their environments.

Welfare and Safety Context for Echolocation Studies

Research involving marine mammals must consider animal welfare and researcher safety. The following considerations apply to echolocation studies.

Minimizing Disturbance

Passive acoustic monitoring is non-invasive because it records sounds without interacting with animals. However, the deployment and retrieval of recording equipment can disturb marine mammals if not conducted carefully. Researchers should follow institutional animal care guidelines and applicable regulations.

Bycatch and Fisheries Interactions

Echolocation research can inform conservation actions that reduce bycatch. The study of human-dolphin cooperative foraging found that cooperative dolphins gain survival benefits by minimizing spatial overlap with bycatch-prone fisheries [8]. Conservation actions targeting fisher behavior could prevent the erosion of this century-old fishery [8]. Researchers should consider how their findings can support bycatch reduction.

Contaminant Exposure

The melon, which plays a primary role in echolocation, can accumulate contaminants such as heavy metals [14]. These accumulations can alter immune function and stimulate inflammatory responses [14]. Researchers studying echolocation should be aware that the acoustic organ is also a site of contaminant accumulation, with implications for animal health.

Professional Escalation Criteria

Researchers and monitoring professionals should escalate concerns to appropriate authorities when they observe signs of marine mammal distress, unusual mortality events, or illegal activities such as harassment or bycatch. Specific escalation criteria include:

• Detection of stranded or injured marine mammals during fieldwork
• Observation of marine mammals entangled in fishing gear
• Evidence of acoustic disturbance from anthropogenic noise sources
• Unusual changes in echolocation behavior that suggest health problems
• Discovery of contaminant levels that exceed regulatory thresholds

Frequently Asked Questions

What is the difference between echolocation and passive listening in marine mammals?

Echolocation is an active process in which an animal emits sound and analyzes the returning echo to form an acoustic image of its environment. Passive listening involves detecting and interpreting sounds produced by other animals or the environment without emitting sound. Odontocete cetaceans use active echolocation, while pinnipeds rely on enhanced passive listening skills, along with visual and tactile senses, to forage and navigate underwater [7].

Do all whales echolocate?

No. Echolocation is found in odontocetes, the toothed whales, which include dolphins, porpoises, sperm whales, and beaked whales. Baleen whales, the mysticetes, do not echolocate. They use other acoustic signals for communication and may use passive listening for navigation and foraging. The evolutionary success of odontocetes is closely tied to echolocation [9].

Why do seals not echolocate?

Pinnipeds, including seals, do not echolocate because of constraints imposed by the obligate amphibious functioning of their auditory system [7]. These constraints have prevented the evolution of the highly acute, aquatic, high-frequency sound production and reception systems required for underwater echolocation [7]. Instead, pinnipeds have evolved enhanced visual, tactile, and passive listening skills [7].

How do dolphins use echolocation to find food?

Dolphins emit echolocation clicks and analyze the returning echoes to detect and discriminate objects, including prey. The echoes provide information about the size, shape, material, and internal structure of objects [12]. Dolphins can discriminate objects that differ only in material composition, as demonstrated in cross-modal matching experiments [12]. During foraging, dolphins adjust their echolocation use based on the behavioral context, such as coordinating with fishers during cooperative foraging [8].

What is the effective detection area in passive acoustic monitoring?

The effective detection area (EDA) is the area around each hydrophone receiver over which vocalizations are detected [3]. Estimating the EDA is necessary to convert click detections into population density estimates [3]. The EDA can be modeled using the passive sonar equation, but common simplifying assumptions about click spectra can bias density estimates [3].

How do researchers classify dolphin echolocation clicks?

Researchers use automated classification methods that incorporate multiple click characteristics, including spectral shape and inter-click interval distributions [13]. Deep learning approaches, including Convolutional Neural

Related Articles

• Marine Biology Courses
• Hydrolyzed Protein: What It Means in Biology, Food, and Laboratory Use
• Animal Behavior Specialist Careers: Veterinary Behaviorists and Trainers
• DVM Programs Compared: US, Caribbean, and International Veterinary Schools
• Exotic, Zoo, and Wildlife Veterinary Careers: Pathways and Opportunities

References and Further Reading

• NCBI Literature Resources. National Center for Biotechnology Information.
• PubMed. National Library of Medicine.
• Modelling the broadband propagation of marine mammal echolocation clicks for click-based population density estimates.. The Journal of the Acoustical Society of America, 2018.
• Performance metrics for marine mammal signal detection and classification.. The Journal of the Acoustical Society of America, 2022.
• Bio-Inspired Covert Active Sonar Strategy.. Sensors (Basel, Switzerland), 2018.
• MAMMALS IN PORTUGAL: A data set of terrestrial, volant, and marine mammal occurrences in Portugal.. Ecology, 2022.
• Why pinnipeds don't echolocate.. The Journal of the Acoustical Society of America, 2000.
• Foraging synchrony drives resilience in human-dolphin mutualism.. Proceedings of the National Academy of Sciences of the United States of America, 2023.
• Convergent evolution in toothed whale cochleae.. BMC evolutionary biology, 2019.
• Using on-board sound recordings to infer behaviour of free-moving wild animals.. The Journal of experimental biology, 2019.
• The application of neural networks to classify dolphin echolocation clicks. 2022.
• Possible limitations of dolphin echolocation: a simulation study based on a cross-modal matching experiment.. 2021.
• Automated classification of dolphin echolocation click types from the Gulf of Mexico.. 2017.
• Involvement of the striped dolphin (Stenella coeruleoalba) melon in immune function: A histological and immunohistochemical study.. 2026.
• Advancing Dolphin Acoustic Monitoring: A Comprehensive Whistle Classification Framework. 2026.
• Open-source machine learning BANTER acoustic classification of beaked whale echolocation pulses. Ecological Informatics, 2024.
• A survey of detection techniques for sperm whale and dolphin echolocation clicks. Frontiers in Marine Science, 2025.
• The passive recording of the click trains of a beluga whale (Delphinapterus leucas) and the subsequent creation of a bio-inspired echolocation model. Bioinspiration & Biomimetics, 2024.
• From looking to echolocation. Posthumanist rhetorics of marine mammal welfare. Res Rhetorica, 2026.
• The mixed blessing of echolocation: Differences in sonar use by fish-eating and mammal-eating killer whales. Animal Behaviour, 1996.
• Echolocation. Encyclopedia of Marine Mammals Third Edition, 2017.
• High frequency echolocation, ear morphology, and the marine-freshwater transition: A comparative study of extant and extinct toothed whales. Palaeogeography Palaeoclimatology Palaeoecology, 2014.

This article is educational and does not replace institutional policy, professional advice, or applicable safety and regulatory requirements.