The research, led by Julia Zeh, a former Ph.D. student in Syracuse University’s College of Arts and Sciences and a member of Susan Parks’ Bioacoustics and Behavioral Ecology Lab, provides an unusually detailed view of what singing whales are doing as they vocalize. Humpback song has been studied for decades because of its complexity and cultural transmission, but scientists have traditionally analyzed it primarily as an acoustic phenomenon. By combining sound recordings with movement data, Zeh and her collaborators examined the physical context in which individual song elements are produced.

The team temporarily attached non-invasive suction-cup tags to 16 singing humpback whales between 2018 and 2024. The work took place in the waters off Maui, part of the Hawaiian breeding grounds where male humpbacks gather during winter. The tags, approximately the size of a tablet, recorded the whales’ underwater sounds and movements before detaching naturally. Researchers deployed the devices from small boats using long poles and, in more recent fieldwork, with the assistance of drones. Once recovered, the tags provided synchronized acoustic and behavioral records for analysis.

The recordings captured an extraordinary level of persistence. Some whales repeated their songs continuously for as long as 16 hours, offering researchers extended sequences in which to compare vocal structure with diving behavior. Humpback songs consist of repeated units arranged into phrases and larger sections known as themes. Although songs can change over time and differ among populations, individual males within a breeding population often share recognizable structures. The new analysis showed that recurring themes appeared consistently during particular portions of a whale’s dive rather than being distributed randomly throughout the performance.

This pattern means that a song can act as a kind of acoustic map of a whale’s behavior. Certain themes were associated with the descent, others with periods of changing depth, and still others with the deepest and most stable portions of a dive or the approach toward the surface. Earlier observations had suggested that some themes were more common near the surface, but the new study links the organization of an entire song to the organization of an entire dive cycle. The result is a connection between the temporal structure of vocal communication and the physical demands of life underwater.

The deepest, relatively stable sections of dives may be especially important for understanding how sexual selection shapes humpback song. Themes performed during these periods displayed different levels of variability from themes sung during other parts of the dive. Zeh and her colleagues propose that the most variable sections could be more strongly influenced by sexual selection, potentially allowing males to signal individual quality or distinguish themselves from competitors. In contrast, song elements associated with ascent, descent or changing depth may be more tightly constrained by physiology and environmental conditions.

Singing while diving is likely to impose competing demands on a whale’s body. Humpbacks must manage limited oxygen stores, control buoyancy and maintain movement while producing powerful, repeated sounds. Changes in depth also alter pressure, and the whale’s body must respond as it moves through the water column. The study does not show that every song theme has a single physiological function, but it indicates that the demands of a particular stage of a dive may influence when that theme is produced and how much it varies. This offers a framework for investigating how acoustic displays evolve under both social and physical constraints.

The findings could also transform how scientists monitor whales. Passive acoustic monitoring is already widely used to detect humpback presence, estimate seasonal activity and study population-level changes without approaching the animals. If specific song themes reliably correspond to particular depths or dive stages, recordings made from hydrophones could provide clues about a whale’s position in the water column even when the animal cannot be seen. Such information could improve estimates of normal diving and singing behavior, creating a baseline against which disturbances from vessel traffic, construction or underwater noise can be measured.

It remains uncertain whether the same relationship between song themes and diving occurs in humpback populations elsewhere. Humpback whales in different oceans share broad biological traits, including the use of learned songs during breeding, but their songs evolve independently and can differ substantially between hemispheres. Physiological limits on oxygen use and buoyancy may be widespread, while local environments, social conditions and cultural traditions could produce different acoustic patterns. Testing the idea will require comparable tag deployments on breeding grounds around the world, where researchers can determine which aspects of the song–dive relationship are universal and which are population-specific.

Published in Current Biology, the study represents the culmination of years of fieldwork and computational analysis carried out during Zeh’s doctoral research at Syracuse. Its significance extends beyond humpback whales: the approach demonstrates how biologging and bioacoustics can be combined to study complex communication in moving animals. By showing that a whale’s song is tied not only to what it sounds like but also to where the animal is in its dive, the research reframes one of nature’s most famous vocal performances as a dynamic interaction between behavior, physiology, environment and evolution.

Subject of Research: The relationship between humpback whale song structure, diving behavior, movement and physiological constraints.

Web References: https://vimeo.com/1214371898/d03d506015?share=copy&fl=cl&fe=ci%C2%A0 ; https://babel.syr.edu/

References: Current Biology; research led by Julia Zeh and collaborators at Syracuse University.

Image Credits: Julia Zeh; NOAA NMFS Permit #19655 and #26593.

Keywords: Humpback whales, whale song, bioacoustics, marine biology, animal communication, diving behavior, passive acoustic monitoring, sexual selection, biologging, ocean noise.

Traditional approaches to acoustic data analysis typically rely on pre-existing knowledge of sound types and manually curated libraries. These conventional methodologies, while effective for known species or signal types, are severely limited when faced with the immense diversity and novelty inherent in ecological soundscapes. The newly developed exploratory protocol bypasses this limitation by employing advanced statistical and computational techniques, enabling researchers to identify unknown and potentially cryptic acoustic patterns without prior labeling or reference data.

At the heart of the protocol lies an intricate framework leveraging unsupervised learning algorithms and sophisticated feature extraction methods. By analyzing temporal and spectral characteristics of recorded sounds, the approach clusters acoustically similar signals, unveiling meaningful groupings that may correspond to distinct behavioral or ecological phenomena. This granular analysis not only accelerates the processing of massive datasets but also enhances sensitivity to rare or transient acoustic events, which often hold crucial ecological information.

The practical application of this protocol involves its deployment on extensive underwater acoustic recordings, an environment notoriously rich in diverse sound sources but challenging for standard analytic tools. These recordings, often collected via autonomous hydrophones, encompass myriad sound emissions from marine mammals, fish, invertebrates, and abiotic phenomena. Decoding this complex soundscape with the new method provides researchers with unprecedented insights into species presence, interactions, and responses to environmental changes, which are critical for conservation and management efforts.

Key to the method’s success is its adaptability and scalability. Unlike traditional template or supervised classification methods, which require extensive annotated datasets and manual intervention, this exploratory protocol can be applied to any large-scale acoustic dataset, regardless of geographical location or target taxa. This universal applicability markedly enhances its appeal for global ecological studies, where datasets are continually growing in both size and complexity.

Moreover, the statistical rigor embedded within the protocol ensures that discovered sound clusters are not mere artifacts but represent statistically significant structures within the acoustic dataset. This distinction is vital for ensuring scientific validity and reducing false positives that may otherwise lead to erroneous ecological inferences. The integration of rigorous validation steps further solidifies confidence in the resultant classifications.

The development and validation of this protocol were made possible through collaborative efforts integrating expertise from bioacoustics, computational ecology, and applied statistics. Such interdisciplinary integration exemplifies the future direction of ecological research, where complex environmental questions are addressed through the convergence of diverse scientific domains and cutting-edge technology.

Beyond its immediate applications in ecology and conservation, the methodology presents exciting possibilities for other fields reliant on acoustic data, such as underwater robotics, environmental monitoring, and even marine resource management. By automating the exploratory analysis of complex soundscapes, the protocol significantly reduces the time and resource investments needed to extract meaningful data from continuous acoustic streams.

Looking forward, the authors envision further enhancements incorporating machine learning advancements and real-time processing capabilities. These improvements would enable near-instantaneous detection of novel acoustic events, providing critical early-warning systems for environmental disturbances or anthropogenic impacts. Such real-time monitoring could prove indispensable in preserving vulnerable ecosystems amid rapid global change.

Importantly, the open-access nature of the protocol ensures that researchers worldwide can adopt and adapt it for their specific needs. By sharing the underlying algorithms and guidelines, the community is empowered to refine the method continuously, promoting a dynamic evolution aligned with emerging scientific challenges and opportunities.

The introduction of this novel exploratory analysis protocol marks a transformative step in acoustic ecology. It not only expands the frontier of what can be discerned from vast acoustic datasets but also bridges the gap between raw data collection and actionable ecological insight. As environmental pressures intensify, such innovative tools become indispensable in safeguarding biodiversity and fostering a deeper understanding of our natural world.

This protocol’s promise extends beyond academia, offering practical applications in policy-making and environmental stewardship. By enabling more accurate detection and monitoring of species and habitats, decision-makers are better equipped with the evidence necessary to enact effective conservation strategies. Consequently, this work not only advances science but also contributes tangibly to global efforts in environmental sustainability and climate resilience.

In summary, this pioneering study presents a novel, statistically robust, and scalable protocol for the exploratory analysis of unknown sound types within extensive acoustic datasets. It represents a significant leap forward in the analysis of complex soundscapes, particularly within underwater environments, and sets a new benchmark for future ecological acoustic research.

Subject of Research: Underwater acoustics and exploratory acoustic data analysis in ecological research
Article Title: A novel protocol for exploratory analysis of unknown sound-types in large acoustic datasets
News Publication Date: 3-Sep-2025
Web References: http://dx.doi.org/10.1111/2041-210x.70134
Image Credits: Katie Turlington
Keywords: Underwater acoustics