Dynamic programming-based range-depth estimation of sperm whale echolocation clicks using a two-element array

In fall 2009 the University of Washington, Applied Physics Laboratory conducted in collaboration with the Oregon State University, a comprehensive field test of a passive‐acoustic Seaglider along the western shelf‐break of the island of Hawaii. During the 3 week mission, a total of approximately 170 h of broadband acoustic data [194 kHz sampling rate] were collected. The recordings were manually analyzed by an experienced analyst for beaked whale (Ziphiidae), dolphin (Delphinidae), and sperm whale (Physeter macrocephalus) echolocation clicks as well as echo sounder pings emitted by boats in the area. Here we present and discuss first results of these data analysis, which revealed that more than 50% of the recorded files (each of 1‐minute duration) contain bioacoustic signals. Furthermore the recorded data and the results of the manual analysis are used to validate and optimize an automated classifier for odontocete echolocation clicks, which was developed in a collaborative effort with San Diego State University. The algorithm is intended to be implemented on the Seaglider to enable species identification by classifying detected echolocation clicks in (near) real‐time during sea trials. [This work is funded by the Office of Naval Research and the U.S. Navy’s Environmental Readiness Division.]

Seasonal occurrence of sperm whales (Physeter macrocephalus) around Kelvin Seamount in the Sargasso Sea in relation to oceanographic processes

The Littoral Acoustic Demonstration Center (LADC) acquired underwater acoustic data from clicking sperm and beaked whales in the Gulf of Mexico during an experiment in July 2007. Analysis of sperm whale echolocation and coda clicks in the data has been presented at previous ASA meetings. Research emphasis is now on identifying individual beaked whales from their echolocation clicks. Again, clustering methods present one appropriate approach to the automated identification of these whales by computer. Self-organizing maps have the ability to limit the number of clusters automatically, and their successful use with sperm whales encourages the application to beaked whales as well. Results for time, frequency, and wavelet domain data allow comparison among clustering done using different signal representations and help highlight areas of concern in the analysis. As in the case of sperm whales, the beaked whale identifications are verified by comparing to cadence analysis (as presented in an accompanying abstract) and to a limited extent by manual analysis and localization. [Research supported by ONR and SPAWAR.]

Odontocete echolocation clicks are generated by pneumatically driven phonic lips within the nasal passage, and propagated through specialized structures within the forehead. This study investigated the highly derived echolocation structures of the pygmy (Kogia breviceps) and dwarf (K. sima) sperm whales through careful dissections (N = 18 K. breviceps, 6 K. sima) and histological examinations (N = 5 K. breviceps). This study is the first to show that the entire kogiid sound production and transmission pathway is acted upon by complex facial muscles (likely derivations of the m. maxillonasolabialis). Muscles appear capable of tensing and separating the solitary pair of phonic lips, which would control echolocation click frequencies. The phonic lips are enveloped by the "vocal cap," a morphologically complex, connective tissue structure unique to kogiids. Extensive facial muscles appear to control the position of this structure and its spatial relationship to the phonic lips. The vocal cap's numerous air crypts suggest that it may reflect sounds. Muscles encircling the connective tissue case that surrounds the spermaceti organ may change its shape and/or internal pressure. These actions may influence the acoustic energy transmitted from the phonic lips, through this lipid body, to the melon. Facial and rostral muscles act upon the length of the melon, suggesting that the sound "beam" can be focused as it travels through the melon and into the environment. This study suggests that the kogiid echolocation system is highly tunable. Future acoustic studies are required to test these hypotheses and gain further insight into the kogiid echolocation system.

AimUnderstanding cetacean species' distributions and population structure over space and time is necessary for effective conservation and management. Geographic differences in acoustic signals may provide a line of evidence for population‐level discrimination in some cetacean species. We use acoustic recordings collected over broad spatial and temporal scales to investigate whether global variability in echolocation click peak frequency could elucidate population structure in Blainville's beaked whale (Mesoplodon densirostris), a cryptic species well‐studied acoustically.LocationNorth Pacific, Western North Atlantic and Gulf of Mexico.Time period2004–2021.Major taxa studiedBlainville's beaked whale.MethodsPassive acoustic data were collected at 76 sites and 150 cumulative years of data were analysed to extract beaked whale echolocation clicks. Using an automated detector and subsequent weighted network clustering on spectral content and interclick interval of clicks, we determined the properties of a primary cluster of clicks with similar characteristics per site. These were compared within regions and across ocean basins and evaluated for suitability as population‐level indicators.ResultsSpectral averages obtained from primary clusters of echolocation clicks identified at each site were similar in overall shape but varied in peak frequency by up to 8 kHz. We identified a latitudinal cline, with higher peak frequencies occurring in lower latitudes.Main conclusionsIt may be possible to acoustically delineate populations of Blainville's beaked whales. The documented negative correlation between signal peak frequency and latitude could relate to body size. Body size has been shown to influence signal frequency, with lower frequencies produced by larger animals, which are subsequently more common in higher latitudes for some species, although data are lacking to adequately investigate this for beaked whales. Prey size and depth may shape frequency content of echolocation signals, and larger prey items may occur in higher latitudes, resulting in lower signal frequencies of their predators.

We present results of classification of toothed whale echolocation clicks to species for 4 TB of recordings in the development data of 7th Intl. DCLDE Workshop. The data span multiple seasons, years, locations, and instruments. Five species were acoustically identified by analysts, and a sixth category was assigned to echolocation clicks that could not be identified to species by analysts. Methods were developed to identify periods of echolocation activity taking into account sporadic false positives that occur throughout the data, and to reduce false positives from both anthropogenic and biologic sources. Dense echolocation activity presented particular challenges for noise removal and long term recordings permitted the targeting of regions for noise estimation outside of echolocation encounters. Extracted features consisted of noise normalized cepstral characterizations of spectra, inter-click interval, and -3 dB bandwidth. These features were classified with a Gaussian mixture model (GMM). A robust sco...

Passive acoustic monitoring (PAM) of marine mammals is essential for biodiversity monitoring and communication studies. Tracking animals in situ further enables behavioral investigation. Monitoring from boats, tags, or moorings is limited by the durability of tracking, whereas underwater gliders offer potential for long-term PAM-based tracking. In this paper, we present the design details of backseat driver operations on such a glider. The backseat driver employs acoustic detection, source separation, and angle of arrival estimation of sperm whale (Physeter macrocephalus) echolocation clicks from a four-element hydrophone array mounted on top of a customized Project CETI-SeaExplorer glider. The measurements are used to change the bearing angle of the underwater glider in real time, allowing it to react to the presence of sperm whales and follow them. The glider is a passive system with no radiated noise except when changing buoyancy (a few seconds roughly every hour), and its distance from the whales is above 100m to have minimal impact on whale behavior. We discuss the operation and connectivity of the backseat driver to the SeaExplorer glider's main computer and how estimation results are shared with the user upon surfacing. While long-term whale monitoring is left for future work, our results from a controlled sea experiment demonstrate the backseat driver operation and show performance in terms of response latency to commands, and results from a field trial off Dominica show the backseat driver source separation capabilities. The code and data for this experiment are open source.

Jacobs et al. (2024) recently published a powerful analysis of apparent source levels from acoustically localised sperm whale codas, using it to estimate that the active space of these vocalisations is around 4 km. This is an important step forward in our understanding of how these vocalisations may function, and we congratulate the authors on the rigour of their analysis. However, we have disagreements with the framing and interpretation of these results that we felt were best addressed in a letter.Our motivations for writing are twofold. The first is to correct the record regarding how Jacobs et al. (2024) frame their work. The second is to critically evaluate their apparent claim that a 4 km active space means that codas do not function to communicate between units.First, Jacobs et al. (2024) motivate their hypothesis that the active space of codas is 'on a par' with the acoustic footprint of a social unit by asserting that we have made specific claims about communication over a 'long range' or 'larger distances' (neither are specified) [Introduction, para 5: 'Such a small active space is, in turn, incompatible with the proposed long-range coda communication for broadcasting clan identity beyond the unit (Gero et al., 2016a,b; Rendell and Whitehead, 2003).' and Introduction, para 6: 'Rendell and Whitehead (2003) suggest that identity codas could mediate the social segregation observed between clans by broadcasting a unit's clan identity to other units over larger distances before engaging in a close-proximity social interaction.'], citing a paper of ours (Rendell and Whitehead, 2003). We did not in fact make any claims about long range communication using codas in that paper. If there are passages that could give the impression of the reported claims, we hope Jacobs et al. (2024) can identify them so we can be clearer in our expression going forward. The one field-derived estimate they quote of 600 m (Weilgart and Whitehead, 1997) is <1/6 the active space estimates they present, while in a book published the same year as the cited paper, one of us inferred an active space for codas of ∼2 km (Whitehead, 2003, p. 135), and we were part of a collaboration that documented short range (<400 m) coda exchange interactions and argued that these support a primary functional hypothesis of social bonding within units (Schulz et al., 2008). We have never suggested that codas are used over long (presumably tens of kilometre or more) ranges.Second, a 4 km active space does not preclude codas functioning to communicate between social units or groups, a claim by Jacobs et al. (2024) that we contest [Title: '…information for intra-unit communication'; Abstract: 'We conclude that while sperm whale codas may contain information about clan affiliation, their moderate active space shows that codas are not used for long range acoustic communication between units and clans'; Introduction, para 6: 'Such a small active space is, in turn, incompatible with the proposed long-range coda communication for broadcasting clan identity beyond the unit'; Discussion, para 1: 'Our estimate … strongly supports our hypothesis that codas are used primarily for within-unit communication']. Clearly it matters here what is meant by 'long range' in this context – without specifying, Jacobs et al. (2024) present it as synonymous with communication between units. While we agree that within-unit communication is likely a major function of codas, we cannot set aside another important feature of sperm whale social structure – units of different clan dialect almost never form coordinated groups in our primary study areas in the Southeast Pacific (Rendell and Whitehead, 2003), as well as in the Caribbean, where habitat use is organised on vocal clan lines (Vachon et al., 2022). This is despite the presence of units of different clans in the same area (Whitehead and Rendell, 2004; Vachon et al., 2022), and that, in the Southeast Pacific, groups consist of a temporary accretion of about 2–3 social units over time scales of 2 to a few days, so that units are likely regularly making decisions as to which other units to group with (Whitehead, 2003). How do they do this if they are not using coda dialect information specifically to make decisions about which units to associate with?A likely scenario, based on the way that we ourselves encounter sperm whale groups acoustically in the field, is that the first cue received between units when they come into proximity is the louder echolocation clicks. This tells listeners that sperm whales are present, but only when they approach closer do they come within the active space of their codas. While a unit footprint might reach 4 km across, codas produced by animals on the periphery will be detectable up to 4 km away from the edge of the unit. At that point, both units can decide whether to approach or avoid while they are still up to 4 km from each other, obviating the need for any close interaction. We have used the same equipment and methods in both the Caribbean and Southeast Pacific study areas, and our field experience is of similar active space in both.While codas might have first evolved primarily for bonding within social units, once the vocal signals are emitted, they become public information identity cues. The natural selection consequences of other social units using this information would make them subject to processes of signal evolution in the between-unit context also (for example, by increasing the chances that units of the same dialect group can find each other and benefit from increased protection). There is therefore no reason to approach the question of whether codas function within or between units from an either/or perspective – evolution can operate in both contexts. Pluripotent signals, where one signal structure has multiple functions, are a known feature in animal communication (Hebets et al., 2016) and nothing in the evidence Jacobs et al. (2024) present precludes this. Duets and choruses in birds, for example, have flexible, context-dependent function (Mennill and Vehrencamp, 2008) and we see no reason the same should not be true for sperm whales, where coda exchanges can often be characterised as duets or choruses (Schulz et al., 2008).We therefore conclude that the active space of sperm whale codas allows for communication both within and between social units.

Sperm Whale: Physeter macrocephalus

We implemented Machine Learning (ML) techniques to advance the study of sperm whale (Physeter macrocephalus) bioacoustics. This entailed employing Convolutional Neural Networks (CNNs) to construct an echolocation click detector designed to classify spectrograms generated from sperm whale acoustic data according to the presence or absence of a click. The click detector achieved 99.5% accuracy in classifying 650 spectrograms. The successful application of CNNs to clicks reveals the potential of future studies to train CNN-based architectures to extract finer-scale details from cetacean spectrograms. Long short-term memory and gated recurrent unit recurrent neural networks were trained to perform classification tasks, including (1) “coda type classification” where we obtained 97.5% accuracy in categorizing 23 coda types from a Dominica dataset containing 8,719 codas and 93.6% accuracy in categorizing 43 coda types from an Eastern Tropical Pacific (ETP) dataset with 16,995 codas; (2) “vocal clan classification” where we obtained 95.3% accuracy for two clan classes from Dominica and 93.1% for four ETP clan types; and (3) “individual whale identification” where we obtained 99.4% accuracy using two Dominica sperm whales. These results demonstrate the feasibility of applying ML to sperm whale bioacoustics and establish the validity of constructing neural networks to learn meaningful representations of whale vocalizations.

The temporal occurrence of deep diving cetaceans in the Josephine Seamount High Seas Marine Protected Area (JSHSMPA), south-west Portugal, was monitored using a passive acoustic recorder. The recorder was deployed on 13 May 2010 at a depth of 814 m during the North Atlantic Treaty Organization Centre for Maritime Research and Experimentation cruise "Sirena10" and recovered on 6 June 2010. The recorder was programmed to record 40 s of data every 2 min. Acoustic data analysis, for the detection and classification of echolocation clicks, was performed using automatic detector/classification systems: M3R (Marine Mammal Monitoring on Navy Ranges), a custom matlab program, and an operator-supervised custom matlab program to assess the classification performance of the detector/classification systems. M3R CS-SVM algorithm contains templates to detect beaked whales, sperm whales, blackfish (pilot and false killer whales), and Risso's dolphins. The detections of each group of odontocetes was monitored as a function of time. Blackfish and Risso's dolphins were detected every day, while beaked whales and sperm whales were detected almost every day. The hourly distribution of detections reveals that blackfish and Risso's dolphins were more active at night, while beaked whales and sperm whales were more active during daylight hours.

Spectral parameters were used to discriminate between echolocation clicks produced by three dolphin species at Palmyra Atoll: melon-headed whales (Peponocephala electra), bottlenose dolphins (Tursiops truncatus) and Gray's spinner dolphins (Stenella longirostris longirostris). Single species acoustic behavior during daytime observations was recorded with a towed hydrophone array sampling at 192 and 480 kHz. Additionally, an autonomous, bottom moored High-frequency Acoustic Recording Package (HARP) collected acoustic data with a sampling rate of 200 kHz. Melon-headed whale echolocation clicks had the lowest peak and center frequencies, spinner dolphins had the highest frequencies and bottlenose dolphins were nested in between these two species. Frequency differences were significant. Temporal parameters were not well suited for classification. Feature differences were enhanced by reducing variability within a set of single clicks by calculating mean spectra for groups of clicks. Median peak frequencies of averaged clicks (group size 50) of melon-headed whales ranged between 24.4 and 29.7 kHz, of bottlenose dolphins between 26.7 and 36.7 kHz, and of spinner dolphins between 33.8 and 36.0 kHz. Discriminant function analysis showed the ability to correctly discriminate between 93% of melon-headed whales, 75% of spinner dolphins and 54% of bottlenose dolphins.

Ocean gliders can provide an inexpensive alternative for marine mammal population density studies. Gliders can monitor bigger spatial areas than fixed passive acoustic recorders. It is a low-noise, low-speed platform, easy to set up, maneuver, and transport on land, deploy, and recover. They can be deployed for long periods and report near real-time results through Iridium modem. Furthermore, gliders can sense the environmental conditions of the survey area, which are important for estimating detection distances. The main objective of this work is to evaluate the use of ocean gliders for population density estimation. Current methodologies developed for fixed sensors will be extended to these platforms by employing both simulations and real experimental data. An opportunistic preliminary sea trial conducted in June 2014 allowed for testing of a Slocum glider fitted with an inexpensive acoustic recording system comprising of two hydrophones connected to an off-the-shelf voice recorder installed inside the glider. Acoustic data recorded in deep waters (>1500 m) off the western coast of Sardinia, Mediterranean Sea, showed the presence of sperm whale echolocation clicks. An improved experiment is planned for the summer 2015. Preliminary results of both campaigns will be presented with an emphasis on population density estimation.

Differences in foraging activity of deep sea diving odontocetes in the Ligurian Sea as determined by passive acoustic recorders

Several beaked whale species occur off eastern Canada. However, except for the northern bottlenose whale (NBW; Hyperoodon ampullatus), their distribution and annual occurrence remain largely unknown, which complicates management efforts to assess the status of poorly known species and effectively protect those species considered at risk. The main objective of this paper is to provide a year-round and pluriannual description of the minimum acoustic occurrence of the NBW, Sowerby’s (SBW; Mesoplodon bidens), Cuvier’s (CBW; Ziphius cavirostris), True’s (TBW; M. mirus) and Gervais’ (GBW; M. europaeus) beaked whales. Twenty-five acoustic recorders were deployed off eastern Canada between May 2015 and November 2017. Beaked whale echolocation clicks were detected using a combination of automated detectors and manual validation at 12 of these stations. Detections were generally restricted to deep continental slope waters. All detected species occurred in the southern part of the study area (off the Scotian Shelf and southern Grand Banks), while only NBWs were detected at the northern edge, off southern Labrador. Clicks identified as TBW or GBW were restricted to, but occurred annually in, the southern areas. All other species were present, at least seasonally, east and north of the Grand Banks. NBWs occurred every day in the Gully Canyon, where SBWs also occurred regularly. While these results should be interpreted as minimum species presence and considered with regards to detector performance, they provide important information regarding beaked whales’ use of areas off eastern Canada where these species have generally received no or very limited monitoring effort.