Toothed whales use sonar to detect, locate, and track prey. They adjust emitted sound intensity, auditory sensitivity and click rate to target range, and terminate prey pursuits with high-repetition-rate, low-intensity buzzes. However, their narrow acoustic field of view (FOV) is considered stable throughout target approach, which could facilitate prey escape at close-range. Here, we show that, like some bats, harbour porpoises can broaden their biosonar beam during the terminal phase of attack but, unlike bats, maintain the ability to change beamwidth within this phase. Based on video, MRI, and acoustic-tag recordings, we propose this flexibility is modulated by the melon and implemented to accommodate dynamic spatial relationships with prey and acoustic complexity of surroundings. Despite independent evolution and different means of sound generation and transmission, whales and bats adaptively change their FOV, suggesting that beamwidth flexibility has been an important driver in the evolution of echolocation for prey tracking.https://doi.org/10.7554/eLife.05651.001
Bats and toothed whales such as porpoises have independently evolved the same solution for hunting prey when it is hard to see. Bats hunt in the dark with little light to allow them to see the insects they chase. Porpoises hunt in murky water where different ocean environments can quickly obscure fish from view. So, both bats and porpoises evolved to emit a beam of sound and then track their prey based on the echoes of that sound bouncing off the prey and other objects. This process is called echolocation.
A narrow beam of sound can help a porpoise or bat track distant prey. But as either animal closes in on its prey such a narrow sound beam can be a disadvantage because prey can easily escape to one side. Scientists recently found that bats can widen their sound beam as they close in on prey by changing the frequency—or pitch—of the signal they emit or by adjusting how they open their mouth.
Porpoises, by contrast, create their echolocation clicks by forcing air through a structure in their blowhole called the phonic lips. The sound is transmitted through a fatty structure on the front of their head known as the melon, which gives these animals their characteristic round-headed look, before being transmitted into the sea. Porpoises would also likely benefit from widening their echolocation beam as they approach prey, but it was not clear if and how they could do this.
Wisniewska et al. used 48 tightly spaced underwater microphones to record the clicks emitted by three captive porpoises as they approached a target or a fish. This revealed that in the last stage of their approach, the porpoises could triple the area their sound beam covered, giving them a ‘wide angle view’ as they closed in. This widening of the sound beam occurred during a very rapid series of echolocation signals called a buzz, which porpoises and bats perform at the end of a pursuit. Unlike bats, porpoises are able to continue to change the width of their sound beam throughout the buzz.
Wisniewska et al. also present a video that shows that the shape of the porpoise's melon changes rapidly during a buzz, which may explain the widening beam. Furthermore, images obtained using a technique called magnetic resonance imaging (MRI) revealed that a porpoise has a network of facial muscles that are capable of producing these beam-widening melon distortions.
As both bats and porpoises have evolved the capability to adjust the width of their sound beam, this ability is likely to be crucial for hunting effectively using echolocation.https://doi.org/10.7554/eLife.05651.002
Echolocation has evolved independently multiple times in mammals and birds (Kellogg et al., 1953; Griffin, 1958; Konishi and Knudsen, 1979; Griffin and Thompson, 1982), and allows these animals to orient under conditions of poor lighting. However, only toothed whales (henceforth ‘whales’) and laryngeally echolocating bats (henceforth ‘bats’) use echolocation to detect, locate, and track prey. In these groups, echolocation signals are primarily ultrasonic (>20 kHz) and are among the most intense biological sounds in water and in air (Madsen and Surlykke, 2013). These characteristics mean that in uncluttered spaces these predators can detect small prey many body-lengths ahead of them. Key to increasing the effective sensory range of biosonar is a directional sound beam (i.e., a narrow volume of forwardly ensonified space), which increases intensity along the acoustic axis and reduces ensonification of objects off-axis. It has recently been proposed that the advantages of a narrow sonar beam while in search of prey may have been the primary driver for the evolution of high frequency sonar signals in both whales and bats (Koblitz et al., 2012; Jakobsen et al., 2013; Madsen and Surlykke, 2013).
As whales and bats close in on prey, both are thought to concurrently decrease signal intensity and auditory sensitivity to partially compensate for reduced transmission loss (Hartley, 1992; Au and Benoit-Bird, 2003; Supin et al., 2004; Linnenschmidt et al., 2012; Madsen and Surlykke, 2013), while increasing the signal emission rate for faster updates on prey location (Griffin, 1958; Morozov et al., 1972; Au, 1993). Both groups terminate prey pursuits with a low intensity, high repetition rate sequence of echolocation signals called a buzz (see ref. [Madsen and Surlykke, 2013] for review). Signal production rates characteristic of buzzes have likely evolved in these echolocators to facilitate close range prey tracking. However, for a given sonar beam, effective beam diameter will decrease as the distance between predator and prey diminishes. Thus, while a directional sound beam enables longer detection range by restricting the acoustic field of view (FOV), it may be disadvantageous to the echolocator at short ranges, as moving prey may easily vanish at the periphery of the FOV. Directionality increases with aperture size (e.g., a bat's gape) and signal frequency (Au, 1993) and both factors appear to be exploited by some bats to modify their beam (Surlykke et al., 2009; Jakobsen and Surlykke, 2010).
Unlike bats, whales do not generate echolocation signals in the larynx nor do they emit them through the mouth or the nares (Ridgway et al., 1980; Cranford et al., 1996). Instead, whales have evolved specialized sound-producing structures, the phonic lips, located high in the blowhole (Figure 1). An echolocation click is generated by pneumatic actuation of the phonic lips as pressurized air is forced past them (Cranford et al., 1996). The resulting sound pulse propagates into the fatty melon and is transmitted into the water as a directional beam (Au et al., 1986). While the exact function of the melon is not known, its size and properties are expected to affect the radiation pattern of sound from the head (Varanasi et al., 1975; Aroyan et al., 1992; Harper et al., 2008). As the bulbous melon fills a large proportion of the forehead (Figure 1), the diameter of the head has come to be considered indicative of radiating aperture size (Au et al., 1999).
Compared to bats, whales exhibit little flexibility in the frequency content of their echolocation signals (Madsen and Surlykke, 2013). We might therefore assume whales are capable of only small changes in beam directionality (Au et al., 1995) and thus be limited to a rather fixed FOV (Au et al., 1999). Yet, given that whales echolocate prey over much longer ranges than bats (Madsen and Surlykke, 2013) and are capable of making much narrower beams (Au et al., 1999; Surlykke et al., 2009; Koblitz et al., 2012; Jakobsen et al., 2013), such a constraint is hard to reconcile with the disadvantages associated with a fixed beamwidth during close tracking of prey. In bats, pursuit is often over quickly, often with less than 500 ms elapsing between prey detection and interception (Kalko, 1995). Prey pursuits by whales can last many seconds (Johnson et al., 2004) (Figure 2). Accordingly, we hypothesize that long buzzes in which the whale might follow its target from an uncluttered water column to a highly cluttered sea floor (Figure 2), and back again, may demand greater beam plasticity than is found in bats.
Corroborating this hypothesis, most of the reported estimates of toothed whale beam patterns show relatively large variability (Evans, 1973; Au et al., 1986, 1987, 1995; Koblitz et al., 2012). Furthermore, a bottlenose dolphin was recently observed to steer, and modify the width of, its sonar beam when stationed on a bite plate and presented with targets displaced by large angles with respect to its body axis (Moore et al., 2008). The dolphin was proposed to use two mechanisms as means of beamwidth modulation: (i) phase shifting between two pairs of phonic lips dually actuated for generation of a single click and/or (ii) manipulation of the volume and geometry of the melon and the associated air-sacs. However, the latest experimental data suggest that dolphins use a single pair of phonic lips to produce echolocation signals (Au et al., 2012; Madsen et al., 2013; Finneran et al., 2014) and that the strong amplitude dependence of dolphin click spectra gives rise to a variable beam pattern, with more directional signals at higher source levels (Finneran et al., 2014). Hence, it remains unclear how and under what circumstances a bottlenose dolphin modulates the width of its sound beam, and whether other whale species, with more stereotyped sonar signals, such as porpoises, are capable of similar beamwidth changes.
In this study, we set out to test the hypothesis that whales can change their FOV adaptively during prey interception. To that end, we used two complementary experiments to record the beam pattern and signal frequency content of echolocation clicks from harbour porpoises (Phocoena phocoena) closing in on targets. Additionally, we obtained magnetic resonance imaging on a dead harbour porpoise to visualize the sound generating structures, the fatty melon and the associated musculature and present video demonstrating melon deformations in an actively echolocating harbour porpoise.
In experiment 1, three captive porpoises were recorded individually using a linear horizontal array of 8 hydrophones spaced 60 cm apart and submerged 75 cm below the water's surface, as the porpoises captured dead fish (Video 1). This setup (Figure 3A) allowed us to quantify the animals' beam patterns over long ranges as they approached and intercepted natural targets free-floating in the water column. A total of 16 trials were recorded for the three animals (5–6 trials/porpoise). Only trials in which the porpoises swam directly towards the centre of the array were analyzed, amounting to two trials for Freja and Eigil and one trial for Sif (a total of 75 clicks). Data from the two porpoises, Freja and Eigil, recorded during buzzes show that they could up to triple the width of their sonar beam as they approached prey (median difference of 7.85° for four trials: from a mean of 12.9° (6.6°–30.5°, with the large beamwidths produced at short target ranges, Figure 3B) during regular clicking (ICI >13 ms; [Wisniewska et al., 2012]) to 19.3° (10.2°–28.9°) during buzzes (ICI ≤13 ms; [Wisniewska et al., 2012]), Figure 3B). However, limitations in the linear array recordings prevented us from drawing strong conclusions about the exact extent of beam widening: the angular resolution was 4–12° and there was no way to precisely determine the vertical direction of the porpoise beam. Even though the video observations indicated that the vertical direction of the porpoise head was not changing with range to the array, this could not be adequately quantified.
We therefore designed experiment two to measure the beamwidth at close target ranges using a non-uniform 2D hydrophone array. One of the three porpoises, Freja, was trained to approach a stationary target surrounded by a star-shaped array of 48 hydrophones in two configurations (Video 2). Initially, the hydrophones were arranged at increasing intervals away from the array center to cover the full extent of the beam at relatively long ranges while providing more resolution around the target at short ranges (Figures 3C, 4, Figure 4—figure supplement 1A). We then repeated the experiment using an array of more-tightly spaced hydrophones (Figures 3C, 4, Figure 4—figure supplement 1B) and suspended the target further in front of the array, to increase signal intensity resolution at short ranges.
We recorded 123 trials with the long-range configuration and 93 trials with the short-range setup. Only trials in which the porpoise swam directly toward the target (within ±15° vertically and horizontally from the array's center) and repeatedly scanned its beam over the center hydrophone were analyzed. This resulted in 11 trials for the long-range- and 4 trials for the short-range array configuration. For each recorded click, the energy levels received at the hydrophones were fitted to the beam pattern of a circular piston (Au et al., 1987; Møhl et al., 2003). Only beamwidth estimates based on fits with R2 > 0.8 (see Figure 4—figure supplement 1) were considered in the analysis, rendering a total of 34 and 458 clicks for the long-range- and short-range array, respectively. The results of this experiment show that the porpoise could almost double its beamwidth in degrees when switching to a buzz (ICI ≤13 ms; [Wisniewska et al., 2012]) at short target ranges (Figure 3D). That is, from a mean half-power beamwidth (i.e., the off-axis angle at which the sound energy has decreased by 3 dB relative to the on-axis energy) of 9.1° (6.7–12.3) to a maximum of 15.1° (mean = 11.0, min = 8.2). In this way the animal increased the ensonified area ahead of it by up to three times (200%) in the terminal buzz phase of its approach to a target, as indicated by the ratio of the area measured at a given target range to the area predicted for that range based on the median −3 dB beam angle calculated for long ranges (i.e., predicted area had the porpoise not switched to ‘wide-angle view’, Figure 5). A significant inverse relationship between the beamwidth of the clicks and distance they were emitted from the target was found for data sets from each experiment (Experiment 1: F = 56.9, R2 = 0.44, p < 0.001; Experiment 2: F = 23.6, R2 = 0.05, p < 0.001). Data from experiment two were also analyzed using ANOVA (see ‘Materials and methods’). The three clusters (sorted using beamwidth values) differed significantly with respect to distance to target (F = 5.21, p < 0.006). Specifically, clicks in the group with the greatest beamwidths were emitted significantly closer to the target than clicks in the group with the narrowest beamwidths (Tukey-Karmer HSD, p < 0.01).
The timeline of these changes (Figure 6A,C) shows that the porpoise varied the width of its beam within the buzz independently of the inter-click intervals (ICIs), with both narrow and wide beam angles used at the lowest ICIs (Figure 6A,D). Consequently, the short- and long-range datasets have different distributions, but similar means, and the long-range clicks are better described by the mean beamwidth value (Figure 3D). The click centroid frequency (fc) dropped by about 1% at the start of a buzz (from a long-range median of 130.4 kHz (127–136.1) to a median of 127.7 kHz (124.1–134.3), see the color scale in Figure 6 and Figure 6—figure supplement 1).
This negligible drop in frequency cannot explain the large changes in beamwidth (Figure 6—figure supplement 1; Spearman's correlation tests between centroid frequency and beamwidth: Experiment 1: p = 0.76, Experiment 2: p = 0.67). Rather, the animal must vary the size of the effective aperture, likely by effecting rapid muscular deformations of the melon (Video 3), the position of the phonic lips and/or the size and position of the associated air sacs. To visualize the musculature surrounding the melon, we obtained magnetic resonance imaging (MRI) on a dead juvenile harbour porpoise. The scanning images reveal nasal structures that include a complex, richly innervated (Huggenberger et al., 2009) network of facial muscles (Figure 1). These muscles are homologous to the muscles dedicated to facial expressions in primates, and should enable fast and subtle changes in the shape of nasal components and their associated air sacs (Huggenberger et al., 2009).
Here, we show that harbour porpoises can broaden their biosonar beams in the final phase of target approach (Figure 3), and, unlike echolocating bats, that they are also able to change their beamwidth within the terminal buzz (Figure 6). At its broadest, the beamwidth ranged from 15° in experiment two to more than 30° in experiment one. A number of factors may have contributed to this variability including: (i) differences in effective angular resolution (EAR) between the eight-hydrophone linear array and the 48-hydrophone star-shaped array, (ii) differences in the animals' behaviour when approaching a slowly sinking fish vs a stationary aluminum sphere, and, finally, (iii) potentially higher clutter originating from the large array behind the target and motivating the porpoise to use a narrower beam. Porpoises can adjust their buzz clicking rate to prey range differently when following a fish in open water than when tracking one in the cluttered and more restricted space close to the sea floor (Figure 2). The porpoise behaviour and experimental context seem the most likely explanation for the observed differences between the two experiments.
This is the first demonstration, to our knowledge, of a whale controlling its acoustic FOV while actively approaching a target. This control is achieved independently of spectral adjustments (Figure 6 and Figure 6—figure supplement 1), probably by changing the conformation of the melon (Figure 1; [Huggenberger et al., 2009]), the position of the phonic lips (Cranford et al., 2014) and the size and shape of the associated air sacs (Aroyan et al., 1992). Due to its heterogeneous structure (Norris and Harvey, 1974; Varanasi et al., 1975), the melon has long been considered an acoustic impedance matcher that minimizes the reflections and energy loss at the tissue–water interface (Norris and Harvey, 1974) and that provides directionality in the emitted click (Au et al., 1999, 2006). Porpoises have a complex facial musculature (Figure 1) with nervous innervation with 4.5 times more neurons than human facial muscles (Jacobs and Jensen, 1964), leading to the recent proposition that muscle induced deformations of the nasal soft structures such as the melon may provide means to change the FOV (Huggenberger et al., 2009). Our acoustic measurements and observations support that hypothesis by demonstrating that the melon and accessory structures apparently operate as the functional equivalent of an adjustable collimating lens of a flashlight. In other words, the porpoise's beam can be dynamically changed from spotlight to floodlight (and everything in between) to best suit the circumstances, offering unprecedented flexibility in control of the FOV in an echolocating animal that is unmatched in visual mammals (Land and Nilsson, 2012).
The porpoise's ability to change its FOV within a buzz (Figure 6) implies an even greater flexibility than recently documented in vespertilionid bats (Surlykke et al., 2009; Jakobsen and Surlykke, 2010; Jakobsen et al., 2013). While bats adjust their beams to the environment in which they are operating (Surlykke et al., 2009), and the task at hand (Jakobsen and Surlykke, 2010), changes in their FOV during the buzz are accounted for by a concurrent drop in signal frequency content. Thus in bats, changes in FOV, emission rate, and signal frequency content during the buzz appear to be tightly interconnected (Ratcliffe et al., 2013). Our results show that in whales these parameters are independently controlled (Figure 6 and Figure 6—figure supplement 1). Having a FOV that can be modulated independently of signal emission rate or frequency content (the latter often being dependent on the amplitude of the signal [Au et al., 1995; Finneran et al., 2014]) may be essential for managing flow of sensory information and optimizing long duration close-range prey tracking in acoustic scenes of varying complexity (Figure 2). The broad beam is advantageous to the porpoises at close range where it would reduce the likelihood of prey escaping perpendicularly to the approaching porpoise by vanishing from its acoustic FOV.
These findings support the hypothesis that porpoises dynamically control their acoustic FOV while tracking prey, and do so by altering the effective size of their radiating aperture. The mechanism underlying these adjustments may be muscle-induced phonic lips repositioning (Cranford et al., 2014), and melon and air sac deformations (Moore et al., 2008; Huggenberger et al., 2009). All toothed whales studied to date have similar facial musculature surrounding the melon (Cranford et al., 1996; Harper et al., 2008). All toothed whales then presumably have the ability to modify the melon's shape. Given the greater beam plasticity offered by this mechanism, compared to modulating the frequency content of clicks, we suggest that all toothed whales may be able to shape and modulate their beam this way.
Despite the independent evolution and very different means of sound generation and transmission, whales and bats have both evolved mechanisms to change their acoustic FOV while tracking prey. This suggests that beam plasticity has been a key driver in the evolution of echolocation, beyond simple orientation, for improved foraging success. Our results from these small toothed whales suggest that the demands of tracking moving prey over variable distances in complex acoustic environments have favored the evolution of a more sophisticated adjustment mechanism, in which pulse rate and beamwidth can be controlled independently. Compared to bats, the greater dynamic beam plasticity we have observed here likely reflects different sensory and ecological constraints. We propose that dynamic control of acoustic FOV in whales is a mechanism for the inclusion and exclusion of potential sensory information that allows these predators to quickly and repeatedly adjust to changes in habitat and prey trajectories.
All experiments were conducted in a semi-natural outdoor enclosure at Fjord&Bælt, situated in Kerteminde harbour, Denmark. The enclosure (approximately 34 × 17 m, natural sandy bottom at 3–5 m depth) is fenced off by a concrete wall alongshore, and nets on the two shorter ends. The net pen complex comprises two net-separated pools that allow for isolation of single animals for experimental work. All acoustic recordings in the present study were made in the smaller, 8 × 12 m net pen.
At the time the study was undertaken, the facility housed four harbour porpoises, three of which participated in the experiments: Freja (female, at Fjord&B;ælt since April 1997, estimated to be 1–2 years old at arrival [Lockyer, 2003]), Eigil (male, at Fjord&B;ælt since April 1997, estimated to be 1–2 years old at arrival [Lockyer, 2003]) and Sif (female, at Fjord&B;ælt since July 2004, estimated to be 1-year-old at arrival [Lockyer, 2003]). All animals had extensive previous experience in various echolocation experiments, from being stationed at a target (e.g., [Beedholm and Miller, 2007; Koblitz et al., 2012; Linnenschmidt et al., 2012]) to free-swimming (e.g., [Verfuss et al., 2005; Beedholm and Miller, 2007; DeRuiter et al., 2009; Wisniewska et al., 2012]), as well as carrying a tag (DeRuiter et al., 2009; Wisniewska et al., 2012). The animals were trained to participate in the experiments using operant conditioning and positive reinforcement (Ramirez, 1999).
We recorded echolocation clicks from three blindfolded (i.e., wearing opaque silicone eyecups) porpoises as they swam alone in the 3–4 m deep net-pen across the long side of the pool to capture dead, freshly thawed fish and towards a horizontal linear array of 8 calibrated Reson TC4014 hydrophones spaced 60 cm apart (Figure 3A and Video 1). The array was deployed at a depth of 75 cm and the fish were introduced approx. 3 m away from its centre, giving the array an effective angular resolution of ∼12° at ranges of target interception (EAR = atan[0.6 m/3 m]). Signals were amplified and filtered using a custom-made conditioning box and then simultaneously A/D converted with 16-bit resolution at 500 kHz per channel (National Instruments PXI-6123, Austin, TX). All trials were monitored using a video camera (Profiline CTV7040, Abus, Germany) synchronized with the audio recordings.
Analyses were performed using Matlab (MathWorks, Natick, MA). We localized the animal at the time of each emission using hydrophone arrival-time differences (Madsen and Wahlberg, 2007). Porpoise positions were then verified with the synchronized videos. For each click, the time of maximum sound pressure on each hydrophone was identified, and the energy density of the signal was measured using a window of 30 µs before and 90 µs after the peak of the signal envelope. Such a window corresponds to the duration of a typical porpoise signal. Assuming spherical spreading, the apparent source level (ASL) (Madsen and Wahlberg, 2007; Finneran et al., 2014) was calculated from the recorded level (computed as energy flux density in a fixed window [peak-30 µs—peak +90 µs]) and the range. For a click to be included in the analysis its ratio of signal energy to the immediately preceding noise energy had to exceed 6 dB on all channels, and the maximum ASL should not have occurred on one of the outermost hydrophones. Radiation plots were created by plotting the ASLs against their respective angles relative to the estimated on-axis direction. First, the peak amplitude and angle were adjusted by interpolating between the peak ASL and the ASLs from the two neighbouring hydrophones using Lagrange interpolation (Menne and Hackbarth, 1986). All off-axis levels were then plotted as a function of the off-axis angle. The resulting transmission beam pattern was interpolated to a grid of 0.1°. The circular piston model was used to estimate the directivity index and −3 dB beamwidth of the beam pattern as described in (Møhl et al., 2003).
A single harbour porpoise (Freja) was trained to swim across the short side of the pool and close in on a 50.8 mm-diameter spherical aluminum target, suspended by a nylon line just in front (5–40 cm, depending on array configuration, see below and Figure 3) of the center of an array of 48 small (25 × 10 mm), custom-built hydrophones (Wisniewska et al., 2012) (Video 2). The hydrophones were attached to a mesh of 1 mm-diameter Dyneema string stretched over a 3 × 3.5 m (height by width) metal frame to form a grid of 5 × 5 cm squares. The hydrophones had a flat (±2 dB) frequency response between 100–160 kHz and were connected to a 48 channel conditioning box with 40 dB gain and a fourth order bandpass filter with −3 dB frequencies at 2 kHz and 200 kHz. The hydrophones were sampled continuously during the trials with 16-bit resolution at 500 kHz/channel by three synchronized National Instruments PXIE-6358 boards and streamed to disk, using custom made software developed in LabVIEW, National Instruments (Source code 1).
Differences in array sensitivity due to hydrophone arrangement and attachment were measured after each data collection and corrected during post-processing. The hydrophones were arranged in two star-shaped configurations (Figure 3C and Figure 4—figure supplement 1). To map the beam extent at long ranges (1.3–7 m), we used hydrophone spacing increasing toward the edge hydrophones 1.05–1.13 m from the centre. Consequently, along the vertical and horizontal axes the hydrophones were separated by: 5, 10, 15, 20, 25, and 30 cm. Along the diagonal axes the hydrophones were separated by 14.1, 14.1, 21.2, 28.3, and 35.4 cm. With hydrophones 14 cm apart, the effective angular resolution was ∼6° at 1.3 m from the array's centre. To maintain high spatial resolution at short ranges (≤2 m), we displaced the target outward to 0.4 m from the array and rearranged the hydrophones, resulting in an array extending out to 0.5 m with hydrophones separated by 5, 5, 5, 15, and 25 cm along the vertical and horizontal axes, and by 7.1, 7.1, 14.1, and 21.2 cm along the diagonal axes. Thus, a sound source at 0.55 m from the array (i.e., the shortest range examined) could have had its beamwidth measured to within ∼5°.
We pooled the two data sets together with a range overlap between 1.3–2 m. Data points acquired with the wider spacing when the animal was closer than 1.3 m were discarded, as were data acquired with the fine spacing when the porpoise was more than 2 m away. The porpoise was equipped with a DTAG-3 multi-sensor tag (Johnson and Tyack, 2003; Johnson et al., 2004; Wisniewska et al., 2012) attached with suction cups just behind the blowhole, to allow for measuring the range of the sound source to the target and the array from the difference in time-of-arrival of click–echo pairs. The tag sampled sound with 16-bit resolution at 500 kHz/channel and was synchronized with the array hydrophones (Wisniewska et al., 2012). In all but one (Figure 6C,D) of the analyzed trials, the porpoise was blindfolded with opaque eyecups.
All trials were monitored with a set of GoPro Hero 2 cameras (two on the heads of the trainers and one approximately 1.8 m behind the array; Eye of Mine Action Cameras, Carson, CA) synchronized with the DTAG-3 recordings.
The recorded trials were pre-screened for relatively straight approaches to the array using the videos. All subsequent sound analyses were performed using Matlab. Clicks from the study animal were identified in the DTAG-3 acoustic recordings using a supervised click detector. Spectral cues were used to eliminate occasional misdetections of echoes or signals from other porpoises in the neighbouring pen. Echograms were formed from the sounds recorded with the DTAG-3. Only trials with clear echoes from the target and the array were submitted to further analysis. Clicks from 13–21 key hydrophones of the large- and small star-shaped array were extracted using a supervised click detector with the synchronized timing of the clicks recorded on the animal as an input. Clicks from all the verified channels were then combined into a single template and used for automatic click detection on the remaining channels. For each click, we identified a subset of channels with peak received levels exceeding the rms noise level of the channel by at least 14 dB. We calculated click energy in a fixed window (peak-30 µs—peak +90 µs) on each of the selected channels, fitted a surface to the values using the Matlab ‘gridfit’ function with grid spacing of 0.5 cm and determined the location of the beam axis as the peak of the fitted surface (Figure 4—figure supplement 1, upper panels). We ran a series of computer simulations of our methods applied to virtual sources of known piston sizes to evaluate the influence of (i) an animal's bearing in azimuth and elevation and (ii) beam axis displacement relative to the centre of the array on the beamwidth measurement error. Based on the results of these simulations, we restricted our analysis to only include clicks emitted when the porpoise was swimming directly toward the target (within ±15° vertically and horizontally from the array's centre) and with beam axis within (i) ±12 cm from the array centre for the large array, (ii) ±8 cm for clicks recorded with the small array and produced up to 1 s before the start of buzz, and (iii) ±6 cm for clicks made 1 s before buzz and onwards until the end of the target approach. The estimated error was thereby limited to ≤±0.5°.
The simulations also verified that at these displacements the location of the beam axis could be estimated with an accuracy of ±1.5 cm at the ranges considered in this study. We used hydrophone arrival-time differences to compute the animal's bearing (azimuth and elevation) to the centre of the array.
For each click, we followed the method of Kyhn et al. (2010) to fit the energy levels received at the hydrophones to the beam pattern of a circular piston that fulfilled the following criteria: (i) it was at the same range to the array as the porpoise emitting the click, (ii) it was centered on the estimated beam axis and (iii) it transmitted the click recorded on the hydrophone closest to that axis. Given that the orientation of the porpoise relative to the array was constantly changing as the animal was approaching the target (Video 2), the beam pattern was assumed to be rotationally symmetrical around the acoustic axis. Furthermore, this assumption allowed us to utilize information from all hydrophones fulfilling the signal-to-noise ratio criterion to find the best fitting aperture. We carried out a Monte Carlo simulation using theoretical piston transducers with diameters of 1/3 to 3 times 8.3 cm (i.e., the best fitting vertical equivalent aperture in [Koblitz et al., 2012]) in 0.1 cm increments, and the circular piston model of Au et al. (1987). The diameter of the piston that matched the data best was found by means of a non-linear least-squares method (see Figure 4—figure supplement 1). Only fits with R2 > 0.8 (Figure 4) were kept for final analysis, from which the −3 dB beamwidth and the equivalent piston radius were extracted.
To examine the relative change in the size of ensonified area ahead of the porpoise as it approached a target, we computed the surface area as base of a cone with height equal to target range and an opening angle corresponding to (i) the measured −3 dB beam angle and (ii) median −3 dB beam angle calculated for long ranges.
We pooled beamwidth and distance to target data for all clicks from Experiment 1 for the three porpoises and ran a regression analysis on these data. We then pooled beamwidth and distance to target data for both array configurations from Experiment 2 and ran a regression analysis on these data as well. Additionally, we used hierarchical cluster analysis (centroid, non-standardized) to assign click beamwidth data from Experiment 2 to one of three clusters and compared these clusters using ANOVA with respect to click distance to target. All statistical analyses were carried out using JMP v. 11.2 (SAS Institute, Cary, NC, USA).
To explore the acoustic scene experienced by a toothed whale tracking active prey, we deployed DTAG-3 tags (with the same recording settings as in experiment two) on porpoises involved in pursuit of small (∼15 cm), live trout in the pen complex of Fjord&Belt;, where the animals have access to a natural sandy bottom at 3–4 m depth. This unique setting approximates what this shallow-water predator might encounter in its natural surroundings. The tags recorded the whale's echolocation clicks as well as echoes from the fish and other objects and surfaces in the whale's surroundings (e.g., water surface and sea floor). The recorded sounds were used to form stack plots, or echograms, of sound envelopes synchronized to the outgoing click as in echosounder images (Johnson et al., 2004). These allowed us to follow movements of the echolocator and its prey in the environment (Figure 2). The time delay between the echolocation click and the echo, multiplied by one-half of the sound speed (1500 m/s was assumed) gives the distance to the target. Delays to the surface and bottom echoes approximate the animal's depth and altitude above the sea floor, respectively. Time delays between different echo groups represent their relative proximity.
A video of melon deformations can be found online (Video 3). To visualize the anatomy of the head, a dead specimen was scanned in a 1.5T Siemens Avanto MRI system (Siemens Medical Solutions, Germany). A Flash 3D T1 weighted pulse-sequence with the following parameters was used: TR 14.8 ms, TE 3.38 ms, α = 15°, NEX = 3, spatial resolution = 0.64 × 0.64 × 0.75 mm. Following acquisition, segmentation and modeling were done using Amira 5.3.3 (Visualization Science Group, Germany).
Computer modeling of acoustic beam formation in Delphinus delphisJournal of the Acoustical Society of America 92:2539–2545.https://doi.org/10.1121/1.404424
The sonar of dolphinsNew York: Springer-Verlag.
Modeling the detection range of fish by echolocating bottlenose dolphins and harbor porpoisesJournal of the Acoustical Society of America 121:3954–3962.https://doi.org/10.1121/1.2734487
Dolphin biosonar signals measured at extreme off-axis angles: Insights to sound propagation in the headJournal of the Acoustical Society of America 132:1199–1206.https://doi.org/10.1121/1.4730901
Acoustic radiation from the head of echolocating harbor porpoises (Phocoena phocoena)The Journal of Experimental Biology 209:2726–2733.https://doi.org/10.1242/jeb.02306
Transmission beam pattern and echolocation signals of a harbor porpoise (Phocoena phocoena)Journal of the Acoustical Society of America 106:3699–3705.https://doi.org/10.1121/1.428221
Echolocation transmitting beam of the Atlantic bottlenose dolphinJournal of the Acoustical Society of America 80:688–691.https://doi.org/10.1121/1.394012
Echolocation signals and transmission beam pattern of a false killer whale (Pseudorca crassidens)Journal of the Acoustical Society of America 98:51–59.https://doi.org/10.1121/1.413643
Acoustic behaviour of echolocating porpoises during prey captureJournal of Experimental Biology 212:3100–3107.https://doi.org/10.1242/jeb.030825
Echolocation by marine delphinids and one species of freshwater dolphinJournal of the Acoustical Society of America 54:191–199.https://doi.org/10.1121/1.1913562
High-resolution measurement of a bottlenose dolphin's (Tursiops truncatus) biosonar transmission beam pattern in the horizontal planeJournal of the Acoustical Society of America 136:2025–2038.https://doi.org/10.1121/1.4895682
Listening in the dark: the acoustic orientation of bats and menNew York: Cornell University Press.
Stabilization of perceived echo amplitudes in echolocating bats: II. The acoustic behavior of the big brown bat, Eptesicus fuscus, when tracking moving preyJournal of the Acoustical Society of America 91:1133–1149.https://doi.org/10.1121/1.402640
Functional morphology of the nasal complex in the harbor porpoise (Phocoena phocoena L.)The Anatomical Record 292:902–920.https://doi.org/10.1002/ar.20854
Gross aspects of the brain and a fiber analysis of cranial nerves in the great whaleJournal of Comparative Neurology 123:55–71.https://doi.org/10.1002/cne.901230107
Vespertilionid bats control the width of their biosonar sound beam dynamically during prey pursuitProceedings of the National Academy of Sciences of USA 107:13930–13935.https://doi.org/10.1073/pnas.1006630107
A digital acoutic recording tag for measuring the response of wild marine mammals to soundIEEE Journal of Oceanic Engineering 28:3–12.https://doi.org/10.1109/JOE.2002.808212
Target detection by an echolocating harbor porpoise (Phocoena phocoena)Journal of the Acoustical Society of America 105:2493–2498.https://doi.org/10.1121/1.426951
Asymmetry and dynamics of a narrow sonar beam in an echolocating harbor porpoiseJournal of the Acoustical Society of America 131:2315–2324.https://doi.org/10.1121/1.3683254
Animal eyes (edition 2)Oxford University Press.
Keeping returns optimal: gain control exerted through sensitivity adjustments in the harbour porpoise auditory systemProceedings of the Royal Society B - Biological Sciences 279:2237–2245.https://doi.org/10.1098/rspb.2011.2465
Harbour porpoises (Phocoena phocoena) in the North Atlantic: biological parametersIn: T Haug, G Desportes, GA Vikingsson, L Witting, editors. Harbour porpoises in the North Atlantic. Tromsø, Norway: North Atlantic Marine Mammal Commission Scientific Publications. pp. 71–89.
Recording and quantification of ultrasonic echolocation clicks from free-ranging toothed whalesDeep-Sea Research Part I: Oceanographic Research Papers 54:1421–1444.https://doi.org/10.1016/j.dsr.2007.04.020
Accuracy of distance measurement in the bat Eptesicus fuscus: theoretical aspects and computer simulationsJournal of the Acoustical Society of America 79:386–397.https://doi.org/10.1121/1.393578
Beamwidth control and angular target detection in an echolocating bottlenose dolphin (Tursiops truncatus)Journal of the Acoustical Society of America 124:3324–3332.https://doi.org/10.1121/1.2980453
Tracking frequency of the location signals of dolphins as a function of distance to the targetBiofizika 17:139–145.
Animal training: successful animal management through positive reinforcementChicago, IL: Shedd Aquarium Society.
Electromyographic and pressure events in the nasolaryngeal system of dolphins during sound productionIn: RG Busnel, JF Fish, editors. Animal sonar systems. New York: Plenum Press. pp. 239–250.
The interaction of outgoing echolocation pulses and echoes in the false killer whale's auditory system: Evoked-potential studyJournal of the Acoustical Society of America 115:3218–3225.https://doi.org/10.1121/1.1707088
Echolocating bats emit a highly directional sonar sound beam in the fieldProceedings Biological sciences/The Royal Society 276:853–860.https://doi.org/10.1098/rspb.2008.1505
Spatial orientation in echolocating harbour porpoises (Phocoena phocoena)Journal of Experimental Biology 208:3385–3394.https://doi.org/10.1242/jeb.01786
Acoustic gaze adjustments during active target selection in echolocating porpoisesThe Journal of Experimental Biology 215:4358–4373.https://doi.org/10.1242/jeb.074013
Russ FernaldReviewing Editor; Stanford University, United States
eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.
Thank you for sending your work entitled “Range-dependent flexibility in the acoustic field of view of echolocating whales” for consideration at eLife. Your article has been favorably evaluated by Eve Marder (Senior editor), a Reviewing editor, and three reviewers.
The Reviewing editor and the reviewers discussed their comments before we reached this decision, and the Reviewing editor has assembled the following comments to help you prepare a revised submission.
Comments for authors:
1) Title: Suggestion to change the title for clarity. Possible title: “Range dependent flexibility in the acoustic field of view of echolocating porpoises.” You might include the species name in the title as well.
2) Abstract: The Abstract appears contradictory in stating that “porpoises… unlike echolocating bats, are able to change beamwidth within” the terminal phase of attack, but also that “both whales and bats have evolved mechanisms to change their FOV… for prey tracking.” You need to be more precise if you want to argue that they have converged but are different. In addition, you state “that harbor porpoises broaden their biosonar beam by >50% in the terminal phase”, but that is not a fair summary of the results, showing very little broadening on average. You should say “harbor porpoises can broaden their…”, to report the results objectively.
a) Please provide more information about the animals used including previous experience with echolocation experiments.
b) Explain why you used only a single animal for the second two experiments and analyzed only a small subsample of the data. Please be clearer about the possible effects of this selective process. Was the performance of their subject exceptional? Did they analyze all the analyzable click trains, or only a subset? Did their criterion reflect a behavioral difference (e.g., the data produced by one type of approach to the target being easier to analyze than another type of approach). Since this phenomenon has not been described previously would be inter2preted differently if the animals used a variety of strategies to home in on the target.
c) The MRI images of the structure of the melon and phonic lips are not described in sufficient detail. Please provide a brief description of how these images compare with previous images of this type from harbor porpoises. Is this the first time simulations of production have been done for the harbor porpoise, or were there others? Were they consistent?
d) There is a discrepancy between the text describing “a dramatic broadening of the beam” and the actual data showing almost no effect, when comparing the median beam width. However, is median the best measure? Figure 4 indicates a strong tendency to broaden the beam in the last phase. Wouldn't this be better described by another measure than median? The data are probably not Gaussian, but even so, the average (or “weight”) might illustrate the effect.
e) How were the data pooled to determine medians and distribution? According to array type? Distance? ICI?
f) Please present a more detailed data taking protocol (individual porpoise, distance between target and array, array type and deployment place) into consideration. Figure 5 shows the results from two very different trials suggesting that pooling the data might mask some of the effects. Figure 3–figure supplement 2 show many data at ca. 1.2 m range. Was there anything specific here? In addition, this figure (which should be included instead of e.g. Figure 2, which is not really used and mostly repeats the message from earlier similar echograms) suggests a very interesting pattern with broadest beam width at between 25 and 35 cm target range.
g) Did individual porpoises show the same results? How did the data from the two arrays compare for the porpoise investigated in both series?
h) What is portrayed in Figure 3–figure supplement 3? If the target was centered on the 0,0 point and the vertical axis was the Y-axis, it can be puzzled out but it should be easier. For example, show the location of the target, and say “bearing” and “azimuth” instead of showing two bearings.
i) At the end of the Introduction, the analysis provided does not directly address the hypothesis to be tested, but perhaps this can possibly be corrected by reanalysis of the existing data, and that this could be done in a reasonable time. The most important result described in the Abstract is that “porpoises broaden their sonar beam by >50%” while the results of experiment 2 state: “the porpoise could almost double its beamwidth at short target ranges (Figure 3D) from a median half-power beamwidth of 9° (6.6-11.8) to a maximum of 15° (median = 10.8, min = 7.4).” So, is the difference 50%, as the Abstract says, or 200% as the Results say? I assume that this does not refer to experiment 1, as the authors state “the relatively large hydrophone spacing in this experiment… prevented us from drawing strong conclusions about the exact extent of beam widening.” The comparison in the Results of experiment 2 is inexplicably between a median and a maximum. If you compare the two median values, 9° (6.6-11.8) to 10.8° (7.4-15), I am not at all convinced that there even is a significant difference. If the paper wants to reach conclusions about changes in beamwidth with distance to target or stages of capture, then it needs a valid statistical comparison of the two distributions. Since the goal is to test whether the porpoise changes FOV adaptively during prey interception, then I think the test should be organized using each approach as the basic unit of analysis. Figure 3D merges data from different approaches, and even totally different receiver set-ups. The actual analysis of adaptive changing of FOV during interception should only test for changes in FOV within each approach. The prediction should be that beamwidth expands as porpoise closes on prey.
j) Do porpoises change beamwidth independent of frequency by changing the configuration of the melon and air sacs? The paper several times cites Figure 5 as demonstrating this. While it shows two examples relating ICI, bandwidth, and center frequency in the context of time from start of buzz, this figure does not actually show a clear analysis that beamwidth is independent of frequency. This requires a statistical test of this point, perhaps against a physical model of how beamwidth would change with fc for a simple static transducer. A plot of beamwidth against fc would also help.
Specific linguistic concerns:
This is a bit picky, but beaked whales only decrease intensity as they switch from search to buzz; most of their closing on prey is not accompanied by decrease. There are very limited data on odontocete species with auditory gain control.
I think you need to walk the reader through the echogram. I think only a tiny minority would be able to link what you say in the text to what you see in Figure 2 without help. Where was this recorded? In the pool? What experiment? The figure is beautiful, but the data in this figure (actual distances to bottom and surface) are not used for anything in the Discussion.
Could the large increase in the experiment with linear array be due to the porpoise not focusing on the array (vertically) before closing in?
How did you control for beam axis in the vertical plane?
I suggest not reporting a difference from 9° to 18° as 99% difference. “100% different” indicate “no overlap”, and accordingly 99% would suggest “almost completely different”. Doubling the sonar beam width would be less confusing.
It is not easy to extract the real results when you provide a comparison between the median (at longer ranges) and the one maximum value at short range. If you want to “break records” it would be fairer to refer to a minimum value vs. a maximum value. Then the reader would know you are referring to the extremes of the dataset.
What does “better approximated” mean precisely?
Please explain which numbers you took to calculate the “at least three times increase”. Figure 5, lower part, seems to indicate that the beam can also be very broad before the buzz.
It is hard to see how the different array configurations could have created such a big difference. But a possible correlation to behavior/context could perhaps: either the freely floating fish with the linear array or the large clutter wall from the star-shaped array. It appears easy to test.
I think that the relation between wavelength and size of the lens/melon differs so radically that this analogy is probably not correct in terms of the physics.
I do not believe that the pupil and lens are not able to change FOV as much as you discuss here for echolocation.
Materials and methods:
How can you accurately interpolate to 0.1° if EAR is only good to 12°?
How did arrangement of the array change the sensitivity of the hydrophones?
Is vertical axis tested?
The subsection headed “Live prey capture”: Is this dataset (and Figure 2) relevant for the study? Apparently, the data are not used for anything. The relevance would be obvious if both the array and the tag had recorded a trial in order to reveal possible correlation between changes in the echogram/acoustic scene and beam width, but that is not the case, as far as I can see.
Do you really think Atlantic cod is a good TS for porpoise prey? Most would be much smaller and lower TS.
Figure 5 is very interesting. The two trials are surprisingly different. Why is that? Were these recorded with the linear array or the star shaped? Probably star shaped, but what configuration?
Video 3: The video shows depression in the whole buzz phase. If that is a proxy for broadening the view, it suggests a constant broad beam in the buzz and not variation as suggested by Results and Discussion in this manuscript?
Please clarify what are the supplementary figures reporting and why.https://doi.org/10.7554/eLife.05651.015
- Danuta M Wisniewska
- Christian B Christensen
- John M Ratcliffe
- Mark Johnson
- Peter T Madsen
- Danuta M Wisniewska
- Kristian Beedholm
- Peter T Madsen
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We are grateful to JH Kristensen, JD Hansen, S Hansen, C Eriksson, M Dyndo, and L Jacobsen and the staff at Fjord&Belt for assistance with data collection, and H Lauridsen at Skejby University Hospital for help with MRI scanning. We thank NU Kristiansen, JS Jensen, M Dyndo, and K Ydesen for helping with construction of the recording setup, H-U Schnitzler (Universität Tübingen) for lending recording gear, and T Hurst (Woods Hole Oceanographic Institution) for providing the DTAG-3. WWL Au, AH Bass, MB Fenton, A Surlykke, and L Wiegrebe kindly provided comments that improved the manuscript. The animals are maintained by Fjord&Belt, Denmark, under permits no. SN 343/FY-0014 and 1996-3446-0021 from the Danish Forest and Nature Agency.
Animal experimentation: The animals are maintained by Fjord&Belt, Denmark, under permits no. SN 343/FY-0014 from the Danish Ministry of Food, Agriculture and Fisheries, and 1996-3446-0021 from the Danish Forest and Nature Agency (under the Danish Ministry of the Environment). Their care and all experiments are in strict accordance with the recommendations of the Danish Ministry of Food, Agriculture and Fisheries (issuing the permit to keep the animals), the Danish Ministry of the Environment (permit for catching the animals) and the Danish Council for Experiments on Animals (always contacted for permits when appropriate—but in the case of this study such permit was not required).
- Russ Fernald, Reviewing Editor, Stanford University, United States
© 2015, Wisniewska et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.