Echolocation (marine mammals)

 

Echolocation is the process in which an animal obtains an, assessment of its environment by emitting sounds and listening to echoes as the sound waves reflect off different objects in the environment. In a very general sense any animal that can emit sounds may be able to hear echoes from large obstacles (i.e., humans yelling in a canyon); however, this type of process is not considered echolocation. The term echolocation is reserved for a specialized acoustic adaptation by animals who utilize this capability on a regular basis to forage for prey, navigate, and avoid predators. Therefore, echolocating dolphins are often searching for entities that are considerably smaller than themselves and must make fine discrimination of these objects. Over the eon of time, this specialized adaptation has been continually refined under evolutionary pressures.

Echolocation in bats was suspected as early as 1912, but it was not until 1938 when G. Pierce and D. Griffin (1938) provided evidence of bats emitting ultrasonic pulses using an ultrasonic detector diat die concept began to gain acceptance. Echolocation in dolphins was suspected around 1947 as was evidenced in the personal notes of A. McBride. the first curator of Marine Studio (later Marineland) in Florida (McBride, 1956). However, it was not until 1960 that Kenneth Norris and colleagues performed the first unequivocal demonstration of echolocation in dolphins by placing mbber suction cups over die eyes of an Atlantic botdenose dolphin (Tursiops truncatus) and observing that the animal wasable to swim and avoid various obstacles. Ultrasonic pulses were also detected as the blindfolded dolphin swam and avoided obstacles, including pipes suspended vertically to form a maze.

Since the Norris demonstration, considerable progress has been made in our understanding of the echolocation capabilities of dolphins. Most of die research has been done with the Atlantic bottlenose dolphin, the most common dolphin in captivity. Research in dolphin echolocation can be divided into the following areas: (1) sound production mechanism and propagation in the dolphin’s head, (2) sound reception and auditory capabilities, (3) sound transmission and the characteristics ofecholocation signals, (4) target detection capabilities, (5) target discrimination capabilities, (6) auditory nervous system function and capabilities, and (7) signal processing modeling. This article addresses each of the first six areas, providing the most recent findings in most cases along with some fundamental capabilities.

I. Sound Production Mechanism and Propagation in the Dolphin’s Head

The head of a dolphin shown in Fig. 1 is a very complex structure with unique air sacs and special sound-conducting fats. Once of the most perplexing issues that has eluded researchers since the discovery of echolocation has been the location and mechanism of sound production in dolphins. In the mid-1980s, Ted Cranford began using modern X-ray computer tomography (CT) and magnetic resonance imaging techniques to study the internal structure within a dolphins head. These noninvasive techniques allowed Cranford to study the relative position, shape, and density of various structures in the dolphins head and helped him to conclude that the monkey lip-dorsal bursae (MLDB) region of the dolphin nasal complex was the location of the sound generator. Eventually, Cranford and colleagues were able to obtain high-speed video in 1997, of the phonic lips (previously referred to as the monkey lips) with simultaneous hydrophone observations of echolocation signals. There are two sets of phonic lips, associated with the two nares in the dolphins nasal complex. Cranford and colleagues have obtained additional high-speed video observation of movements in both sets of phonic lips during the production of echolocation signals and whistles.

 (a) Schematic of a dolphin's head (adapted from Norris, 1968) and (b) three-dimensional diagram of the air sacs in a dolphin's head. PS, premaxillanj sac; VS, vestibular sac; NS, nasofrontal (tubular) sac; AS, accessory sac.

Figure 1 (a) Schematic of a dolphin’s head (adapted from Norris, 1968) and (b) three-dimensional diagram of the air sacs in a dolphin’s head. PS, premaxillanj sac; VS, vestibular sac; NS, nasofrontal (tubular) sac; AS, accessory sac.

The numerical simulation of sound propagation in die head of a dolphin by James Aroyan, then a Ph.D. student at die University of California Santa Cruz, has provided considerable understanding of the role of the air sacs, skull, and melon in the propagation of sounds in a dolphins head. One of the interesting problems Aroyan considered was that of a plane wave propagatingtoward the head of a dolphin (as depicted in Fig. 2a) to determine where sounds would focus in the dolphins head in a similar manner to die process used by geologists to determine the epicenter of an earthquake. He numerically solved die three-dimensional wave equation (also shown in Fig. 2) using a finite-difference technique and a super computer. The density and sound velocity structure of the dolphins head were estimated from the CT scan results of Ted Cranford. The grid points represent a pictorial illustration of how the head of a dolphin may be mathematically subdivided so that the solution of die wave equation is numerically determined at each grid point. The results for the geometry depicted in Fig. 2a are shown in Fig. 2b, with focal regions at die two auditory bullas and at the MLDB region of the nasal system, supporting Cranfords earlier suspicion of die MLDB being the site of die sound generator for echolocation sounds.

Aroyan then placed a hypothetical sound source at the MLDB region and numerically solved the three-dimensional wave equation as the sound propagated through the dolphins head into the water. He found that the skull, the various air sacs, and the nonhomogeneous melon all played important roles in forming the beam in which sounds are transmitted into the free field. The specific characteristics of this beam are discussed in Section III. He also showed that if a hypothetical sound source was placed at the larynx, the resulting beam in the free field was not compatible to the actual beam measured for dolphins, therefore essentially eliminating the larynx as a possible site for a sound generator.

(a) Configuration of numerical simulation of sound propagation in the head of a dolphin to determine acoustic focal regions and (b) results of numerical simulation for the geometry depicted.

Figure 2 (a) Configuration of numerical simulation of sound propagation in the head of a dolphin to determine acoustic focal regions and (b) results of numerical simulation for the geometry depicted.

II. Sound Reception and Auditory Capabilities

A. Sound Reception Site

Dolphins do not have pinnae and their external auditory meatus is but a pinhole with vestigial fibrous tissues connecting the surface to the tympanic structure. Kenneth Norris was the first to postulate that sound enters the dolphins auditory system through the thin posterior portion of the mandible (see Fig. la) and is transmitted via a fat-filled canal to the tympano-periotic bone, which contains the middle and inner ears. Two electrophysiological measurements, one by Theodore Bullock and colleagues, were conducted that provided evidence to support Norris’ theory. However, the acoustic conditions for both experiments were less than ideal: the subjects were confined to small holding tanks and their heads were held near the surface to keep the electrodes from shorting out by the water. The acoustic propagation for such a situation can be extremely variable, with the sound pressure level changing drastically because of multipath propagation, on the order of 10 to 20 dB.

Bertel M0I1I and colleagues took a slightly different electrophysiological approach by measuring the brain stem-evoked potential of a bottlenose dolphin that was trained to beach itself on a rubberized mat. A special suction cup hydrophone having a water interface between the piezoelectric element and the skin of the dolphin was positioned at different locations on the dolphin s head. By performing the measurement in air, the point at which sound from the piezoelectric element enters into the dolphin could be firmly established. Acoustic energy will only propagate toward the dolphin s skin, and energy propagating in any other direction will be reflected back at the boundary of the suction cup. M0hl and colleagues positioned the suction cup hydrophone at different locations around the dolphins head, and at each location, the amount of attenuation needed to obtain the evoked potential threshold was determined. Their results are shown in Fig. 3, where the circles indicate the different positions of the suction cup and the number within each circle represents the amount of attenuation needed to achieve threshold. Therefore, a larger number is indicative of a more sensitive region of sound reception. The dashed line indicates the area of the pan bone or mandibular window shown in Fig. la. These results indicated that the area just forward of the pan bone area of the dolphins lower jaw is the most sensitive area of sound reception, which seems to be inconsistent with Norris’ pan bone theory. However, the numerical simulations of acoustic propagation by Aroyan suggest that sounds that enter the dolphin’s lower jaw just forward of the pan bone actually propagate below the skin surface to the pan bone and enter into the lower jaw through the pan bone.

B. Hearing Sensitivity

The hearing sensitivity of a dolphin was first measured accurately in 1967 by Dr. Scott Johnson in a pioneering experiment. Johnson found that the upper limit of hearing of an Atlantic bottlenose dolphin was 150 kHz. Since Johnson’s research, audiograms have been determined for the harbor porpoise (Phocoena phocoena), Amazon River dolphin (Inia geoffrensis), beluga whale (Delphinapterus leucas), false killer whale (Pseudorca crassidens), Chinese river dolphin (Lipotes vexillifer), Risso’s dolphin (Grampus griseus), Tucuxi (Sotalia fhiviatilus), and killer whale (Orcimts orca). The audiograms for these odontocetes are shown in Fig. 4.

One of the most remarkable features of these audiograms is the high upper frequency limit of hearing extending beyond 100 kHz. This is rather remarkable when the wide range of sizes of the animals depicted in Fig. 4 is considered. The largest animal represented in Fig. 4 is the killer whale, which weighed about 3600 kg and was about 5 m in length compared to the smallest animal, the harbor porpoise, which typically weighs about 33 kg and is about 1.3 m in length. The typical rule of thumb in mammalian hearing is that larger animals tend to have limited high-frequency hearing capabilities. A killer whale is considerably larger than the smallest dolphin, yet its upper limit of hearing is approximately 120 kHz, and may therefore represent an exception to the norm. Another interesting feature of Fig. 4 is the fact that the maximum sensitivity for the different odontocetes has a very similar value, within 10 dB.

Relative reception sensitivity of hearing at different locations around a dolphin's head. The higher the number, the more sensitive the region.

Figure 3 Relative reception sensitivity of hearing at different locations around a dolphin’s head. The higher the number, the more sensitive the region.

Audiograms for 10 species of odontocetes.

Figure 4 Audiograms for 10 species of odontocetes.

Although dolphins do not have pinnaes, the auditor system of the dolphins is directional. Sounds are received best when the source is directly in front of an animal. The receiving beam pattern of a T. truncatus in both horizontal and vertical planes is shown in Fig. 5. Several features of the beam patterns are worth pointing out. First, the beam becomes wider as the frequency decreases. Planar transducers also behave in a similar fashion with their beam becoming narrower as the frequency increases. Second, the major axis of the beam in the vertical plane is pointed upward with respect to the teeth line by about 5-10°. Third, the major axis in the horizontal plane is pointing directly in front of the animal parallel to the longitudinal axis of the dolphin. The receiving beam pattern can also be discussed in terms of the spatial variation in the hearing sensitivity of the dolphin. Therefore, the dolphin has the best high-frequency hearing sensitivity when sounds approach from the front and poorer sensitivity as the sound sources move to other locations about the animals head.

III. Sound Transmission and the Characteristics of Echolocation Signals

There is a distinct difference in the echolocation signals used by odontocetes that produce whistle signals and those that do not whistle. Whistling dolphins project short, almost exponentially decaying signals with durations of 40 to 70 jxsec and band-widths of tens of kHz. Nonwhistling dolphins and porpoises project signals with much durations of 120 to 200 jxsec and with narrow bandwidths that are typically less than 10 kHz. An example of a typical echolocation signal produced by an Atlantic bottlenose dolphin is shown in Fig. 6, along with a typical echolocation signal produced by a harbor porpoise (a nonwhistling animal). Whether riverine dolphins produce whistles is still an open question: however, these dolphins emit signals that are of the broadband, short duration variety. Most odontocete species produce whistles, and only a few, such as the harbor porpoise, Commerson’s dolphin (Cephalorhynchus commersonii), Hectors dolphin (C. hectori), Dalls porpoise (Phocoenoides dalli), and pygmy spenn whale (Kogia breviceps), are known to not.

The amplitudes of the echolocation signals also are very different between whistling and nonwhistling odontocetes. Whistling dolphins, such as T. truncatus, P. crassidens, and D. leucas, can project echolocation signals with peak-to-peak amplitudes as high as 225 dB re 1 jxPa. The center frequency of the signals used by whistling dolphins is affected by the level of the outgoing signal. The center frequency of clicks varies almost linearly with the peak-to-peak amplitude. Nonwhistling dolphins and porpoises, such as P. pliocoena and Phocoenoides dalli, emit signals that normally do not exceed 170 dB re 1 jxPa. Peak-to-peak source level measurements for P. phocoena by Au and colleagues in 1999, while the animal was performing a target detection task, indicated an average peak-to-peak source level of only 160 dB, which is considerably smaller than the 210-225 dB used by T. truncatus, P. crassidens, and D. leucas. The center frequency of the P. phocoena signal, which is typically between 120 and 145 kHz, does not depend on the level of the projected sonar signals.

The receiving beam patterns in the horizontal and vertical planes for different frequencies.

Figure 5 The receiving beam patterns in the horizontal and vertical planes for different frequencies.

Typical echolocation signal emitted by T. truncatus (a whistling dolphin) and P. phocoena (a nomvhistling porpoise). The source level (SL) is the peak-to-peak sound pressure level referenced to 1 (j,Pa at 1 m.

Figure 6 Typical echolocation signal emitted by T. truncatus (a whistling dolphin) and P. phocoena (a nomvhistling porpoise). The source level (SL) is the peak-to-peak sound pressure level referenced to 1 (j,Pa at 1 m.

The transmission beam pattern in horizontal and vertical planes. The signals shown with each beam pattern are all the same signal captured simultaneously by five hydrophones located about the dolphin's head.

Figure 7 The transmission beam pattern in horizontal and vertical planes. The signals shown with each beam pattern are all the same signal captured simultaneously by five hydrophones located about the dolphin’s head.

Echolocation signals are projected from a dolphin’s head in a beam. An example of the transmitting beam pattern for a T. truncatus in both horizontal and vertical planes is shown in Fig. 7. The signal shown at different angles about the animal’s head is the same signal captured by an array of hydrophones. Note that only the signal traveling along the major axis of the beam is undistorted. This phenomenon occurs in horizontal and vertical planes. The numbers above each signal are the maxima in the frequency spectra of the signals, in order of descending amplitude. The further away from the major axis of the beam, the more the signal is changed. This property of the beam makes it very difficult to measure echolocation signals in the wild. Occasionally, dolphins in the wild may actually swim directly toward a hydrophone so that relatively true measurements can be made.

Beam pattern measurements have also been conducted for D. leucas, P. crassidens, and P. phocoena. The signals from all of these animals, with the exception of P. phocoena, exhibit changes in frequency content when the measuring hydrophone is located away from the major axis. However, in the case of P. phocoena, the signals detected by hydrophones located away from the major axis are not distorted, as can be seen in Fig. 8. Distortion does not occur because the signal has a relatively narrow bandwidth.

IV. Target Detection Capabilities

One of the most fundamental properties of a sonar system is its maximum detection range. A simple way to determine the maximum target detection range of a sonar is to gradually move a specific target away from the sonar until the target can no longer be detected. An (1993) used a 7.62-diameter water-filled stainless-steel sphere as the target to determine the maximum detection range of T. truncatus. The target was moved progressively away from an echolocating dolphin until the animal could no longer detect its presence. Stringent psychophysical techniques were used and many sessions were conducted in order to stabilize performance and to determine the probability of detection as a function of range. Kastelein and colleagues (1999) used the same type of target (7.62-cm diameter water-filled stainless-steel sphere) as An and colleagues to determine the sonar detection capability of P. phocoena. The results of both experiments are shown in Fig. 9.

The waveform of an echolocation signal detected by hydrophones spaced about the head of a P. phocoena.

Figure 8 The waveform of an echolocation signal detected by hydrophones spaced about the head of a P. phocoena.

The 50% correct detection threshold for the bottlenose dolphin occurred at a range of 113 m. The 50% correct detection threshold for the harbor porpoise was approximately 26 m. An experiment by An and Snyder took place in Kaneohe Bay, Hawaii, a bay that has a high level of snapping shrimp sounds or noise. Therefore, the bottlenose dolphin was masked by the background noise level. The harbor porpoise target detection experiment was performed at a location in the Netherlands, where the ambient background noise, between 100 and 150 kHz was essentially at sea state 0. Therefore, the harbor porpoise was not masked by background noise and yet its detection threshold was considerably shorter than Tursiops’. The difference in the two-way propagation losses for 113 and 26 m is 36 dB. The bottlenose dolphin typically produces clicks that are 50-60 dB greater than that of the harbor porpoise. Therefore, the large difference in the levels of projected signals can account for most, but not all, of the very large difference in the detection threshold ranges of both animals. If the target detection experiment with T. truncatus were conducted in a body of water with a low ambient background noise, the dolphins target detection range would be much longer and, in that case, the difference in the two-way transmission would probably match the difference between the source levels used by the two different species.

Target detection performance as a function of range for T. truncatus and P. phocoena. After Au (1993) and Kastelein (1999).

Figure 9 Target detection performance as a function of range for T. truncatus and P. phocoena. After Au (1993) and Kastelein (1999).

V. Target Discrimination Capabilities

There have been many target discrimination experiments involving echolocating dolphins. Unfortunately, in many of these experiments the reflection characteristics of the targets were not measured or were measured with tone-burst signals instead of with a simulated dolphin-like signal. The experiment involving wall thickness discrimination by an echolocating dolphin is one that provided appropriate echo characteristics of the targets (An, 1993). In this experiment, the dolphin was presented with two hollow aluminum targets separated by 20° at a range of 8 m. The standard target had a wall thickness of 0.63 cm. and the comparison targets had wall thicknesses that were different than the standard by ± 0.8, ± 0.4, ± 0.3, and ± 0.2 mm. All the targets had a length of 12.7 cm. On any given trial, the standard and comparison were introduced into the water separated by ± 20° about the center axis. The dolphin was required to swim into a hoop and echolocate the targets when a screen was lowered out of the way and then touch a paddle that was on the same side of the center line as the standard target. Two sets oi targets were available so that the position of the standard could be switched on any given trial.

The results of the wall thickness experiment are shown in Fig. 10. The dolphin performed very well with correct responses in the mid-90 percentile. The animal’s correct response performance became progressively worse as the difference in wall thickness decreased. The 75% correct performance threshold was 0.27 mm for the case in which the comparison targets were thinner than the standard target and 0.23 mm when the comparison targets were thicker than the standard.

Echoes from the standard target and the —0.3 mm comparison target are shown in Fig. 11. There are several cues that the animal might have used in order to perform this discrimination. One cue is the difference in the time delay between the first and the second echo component for both the standard and the comparison target. If this cue was used, it suggests that the dolphin could discriminate differences of about 0.5-0.6 |xsee.

Performance of an echolocating dolphin in the wall thickness discrimination experiment.

Figure 10 Performance of an echolocating dolphin in the wall thickness discrimination experiment.

Echoes from the standard and the —0.3 mm targets. The top two traces are the echo waveforms, the middle trace is the envelope of the echo waveforms overlaijed upon each other, and the bottom curve is the spectra of the echoes.

Figure 11 Echoes from the standard and the —0.3 mm targets. The top two traces are the echo waveforms, the middle trace is the envelope of the echo waveforms overlaijed upon each other, and the bottom curve is the spectra of the echoes.

Another cue could be the difference in the time-separation pitch (TSP). When humans are presented with two correlated broadband acoustic signal separated by time T, a time-separation pitch equal to 1 IT can be perceived. TSP stimulii also have a frequency spectrum that is rippled. The third possible cue is the difference in the frequency spectra of the echoes. The frequency spectrum of the echo from the —0.3 mm comparison target is shifted between 2 and 3 kHz from the spectrum of the standard target. If the dolphin was using this cue, it suggests that the animal was able to perceive a frequency difference of 2-3 kHz in the broadband echoes.

VI. Auditory Nervous System Response

The fact that echolocating dolphins can produce signals with peak frequencies between 100 and 135 kHz implies a relatively fast nervous system response because the time between periods of sound transmission and the auditory response can be short, on the order 7 to 10 (Jisec. The speed of the auditory response can be determined by performing an electrophysiological experiment in which the time difference between the projection of the stimulus and the onset of the brain stem response is measured. This time difference is typically referred to as the latency of response. A comparison of the latency for a variety of mammals is shown in Fig. 12. In order to fully appreciate Fig. 12, it is important to understand that the brain stem-evoked potential consists of sev eral waves (indicated by Roman numerals) that arrive at the measuring electrodes at slightly different times. From Fig. 12, it is obvious that dolphins have an extremely fast auditor)’ nervous response, even faster than that of a rat. The acoustic stimulus must travel into the inner ear where the cochlea nerves discharge and send the electrical pulses to higher auditory centers and eventually to the brain stem. What is very astonishing is the fact that a rat’s head is considerably smaller than that of dolphins, yet dolphins have shorter responses than the rat.

The response time of the auditory system of a dolphin can be estimated by performing an integration-time experiment. Au (1993) performed a target detection experiment using phantom echoes. The phantom echo generator would digitize the outgoing signal, which was detected by a hydrophone 1.5 m in front of the dolphin stationing in a hoop. The digitized signal was stored in the memory of a personal computer, and at the appropriate time delay representing the two-way transit time for a target at 20 m, the “echo,” was sent back to the dolphin via a small transducer located 2 m in front of the animal. In the initial phase, a single echo was sent to the dolphin and the dolphin’s hearing threshold for the single echo was determined by varying the amplitude of the echo in a staircase fashion. In the second phase, two echoes were separated by a variable spacing that was sent back to the dolphin. The dolphin’s threshold was obtained by varying the amplitude of the whole echo in a staircase fashion for various separation times between the two echoes.

The results of the phantom echo experiment are shown in Fig. 13. For an echo consisting of a single click, the threshold is shown on the ordinate of the curve in Fig. 13. Then two echoes were sent back to the animal with a separation time of 50 fjisec. The threshold for the two-click echo at 50 (Jisec was approximately 3 dB lower than for the single click threshold. This was expected because the two-click echo had twice the amount or 3 dB more energy than the single click echo, and the dolphin auditory system, like most mammals, behaves as an energy detector. As the time separation between the two clicks increased to 200 |xsee, the dolphins threshold remained very constant. However, when the time separation increased to 250 (xsec and beyond, the dolphins threshold began to move toward the threshold for a single click echo. The solid line in Fig. 13 represents the output of an energy detector having an integration of 264 (xsec. The energy detector response with a 264-(xsec integration time best fits the animal results. This integration time is extremely small compared to the integration time of any other mammal.

Brain stem-evoked potential latency for different animals.

Figure 12 Brain stem-evoked potential latency for different animals. 

Results of the phantom echo experiment.

Figure 13 Results of the phantom echo experiment.

VII. Conclusions

Dolphins have very keen echolocation capabilities, much keener than any man-made sonar, especially in a shallow water environment. The use of relatively short broadband echolocation signals by whistling dolphins is probably the most important factor in the dolphin’s good discrimination capabilities. The broad frequency range of hearing extending over 10 octaves and the good peak sensitivity of 30 to 40 dB re 1 |xPa are certainly contributing factors in the dolphin’s echolocation capabilities. Another feature of the dolphin’s auditory system that contributes to its good echolocation capabilities is the extremely rapid response of its auditory nervous system. The auditory nervous system of the dolphin probably responds faster than that of any other animal if the relative dimensions of the auditory system are taken into account. Finally, dolphins are extremely mobile and can investigate objects at different aspects and angles to maximize the amount of echo information from objects and thus enhance their echolocation capabilities.

Next post:

Previous post: