4. Acoustic Communication in Whales and Dolphins
189
Weilgart 1990, 1991). Weilgart and Whitehead (1988) call these "usual
clicks," but we will call them "regular clicks" because of their regular
interclick intervals. This association of regular clicks with diving and feeding
has led most biologists to hypothesize that these regular click trains function for echolocation (Backus and Schevill1966, Gordon 1987, Whitehead
and Weilgart 1990). Watkins (1980a), on the other hand, argued that sperm
whale clicks were not suited to echolocation of prey. For example, the clicks
of sperm whales are lower in frequency, longer in duration, and much less
directional than the high frequency clicks of dolphins (Au 1993). This led
Watkins (1980a) to argue that regular clicks are social signals used by diving
whales to maintain contact with one another.
The echolocation hypothesis for the regular clicks of sperm whales has
not been definitively tested. However, Goold and Jones (1995) have used
acoustic models to evaluate the potential range at which the regular clicks
of sperm whales might be used to detect their squid prey. These calculations
necessarily involve assumptions and rough estimates of some parameters,
but they suggest that sperm whales might detect squid at a range of 200 to
680m. This range is consonant with the low end of interpulse intervals
observed in regular clicks, assuming the whale waits to make a new
click until it hears an echo return. The speed of sound in seawater is near
1,500m/s, so the round trip travel time to a target 750m away would be
about one second. If sperm whales producing regular clicks wait until an
echo returns before producing the next click, then the 0.5 to 2.0 second
intervals between regular clicks would suggest maximum working ranges
of between about 375 and 1,500m. While the shortest intervals correspond
to the estimated detection range, the longer intervals involve much longer
round trip travel times.
Unfortunately, Goold and Jones (1995) did not correct for the bandwidth
of hearing in their application of the sonar equation, and this inflates the
estimated range of detection. The threshold at which sperm whales are
modeled by Goold and Jones as able to detect echoes from their clicks
occurs when the echo level equals the ambient noise level. This is a reasonable simplification, but it is critical to match the noise and echo levels
in a band appropriate for the hearing of the animals. Goold and Jones did
not do this, but they compare a broadband estimate of the source level of
sperm whale clicks across a frequency range of thousands of Hertz to a
spectral level of noise in 1Hz band. Most mammals integrate sound energy
over frequency bands roughly about one-third octave in breadth. Ideally,
one would measure both signal and noise in the biologically relevant bands
when a click is being made. Insufficient data are presented in the Goold
and Jones paper to do this. If we simply correct the spectral noise estimates
used by Goold and Jones to third octave band levels, leaving all other
aspects of the calculation the same, the estimated detection range drops
from 200m to 38m at 2kHz and from 680m to 108m at 10kHz. The simple
correction performed here will still overestimate effective range of the
189
Weilgart 1990, 1991). Weilgart and Whitehead (1988) call these "usual
clicks," but we will call them "regular clicks" because of their regular
interclick intervals. This association of regular clicks with diving and feeding
has led most biologists to hypothesize that these regular click trains function for echolocation (Backus and Schevill1966, Gordon 1987, Whitehead
and Weilgart 1990). Watkins (1980a), on the other hand, argued that sperm
whale clicks were not suited to echolocation of prey. For example, the clicks
of sperm whales are lower in frequency, longer in duration, and much less
directional than the high frequency clicks of dolphins (Au 1993). This led
Watkins (1980a) to argue that regular clicks are social signals used by diving
whales to maintain contact with one another.
The echolocation hypothesis for the regular clicks of sperm whales has
not been definitively tested. However, Goold and Jones (1995) have used
acoustic models to evaluate the potential range at which the regular clicks
of sperm whales might be used to detect their squid prey. These calculations
necessarily involve assumptions and rough estimates of some parameters,
but they suggest that sperm whales might detect squid at a range of 200 to
680m. This range is consonant with the low end of interpulse intervals
observed in regular clicks, assuming the whale waits to make a new
click until it hears an echo return. The speed of sound in seawater is near
1,500m/s, so the round trip travel time to a target 750m away would be
about one second. If sperm whales producing regular clicks wait until an
echo returns before producing the next click, then the 0.5 to 2.0 second
intervals between regular clicks would suggest maximum working ranges
of between about 375 and 1,500m. While the shortest intervals correspond
to the estimated detection range, the longer intervals involve much longer
round trip travel times.
Unfortunately, Goold and Jones (1995) did not correct for the bandwidth
of hearing in their application of the sonar equation, and this inflates the
estimated range of detection. The threshold at which sperm whales are
modeled by Goold and Jones as able to detect echoes from their clicks
occurs when the echo level equals the ambient noise level. This is a reasonable simplification, but it is critical to match the noise and echo levels
in a band appropriate for the hearing of the animals. Goold and Jones did
not do this, but they compare a broadband estimate of the source level of
sperm whale clicks across a frequency range of thousands of Hertz to a
spectral level of noise in 1Hz band. Most mammals integrate sound energy
over frequency bands roughly about one-third octave in breadth. Ideally,
one would measure both signal and noise in the biologically relevant bands
when a click is being made. Insufficient data are presented in the Goold
and Jones paper to do this. If we simply correct the spectral noise estimates
used by Goold and Jones to third octave band levels, leaving all other
aspects of the calculation the same, the estimated detection range drops
from 200m to 38m at 2kHz and from 680m to 108m at 10kHz. The simple
correction performed here will still overestimate effective range of the