3. Impulse Sound Sources
115
same location with homologous structures in different species (Cranford
1992a).1 Exploration of these discrete acoustic differences is correlated with
anatomic differences that suggest distinctions in the physiologic mechanism. Typically, the pulses we associate with echolocation behavior in a few
well-studied dolphin species contain only a few pressure cycles (three to
seven) in the time domain. These "typical" signals have relatively broad
bandwidths per pulse (30 to 60kHz, at -3dB; see Table 7.2 on page 134 in
Au 1993) and may have a bimodal distribution in the frequency spectrum
(Au et al. 1995; Sigurdson 1997a,b).
By contrast, the pulses produced by porpoises (Phocoenidae) and some
small delphinids, such as Cephalorhynchus commersonii (Commerson's
dolphin), contain many cycles (8 to 20) per pulse and have a narrow
bandwidth character «25 kHz at -3 dB) with a unimodal distribution in
the frequency spectrum (Dubrovskiy et al. 1971; M0hl and Andersen 1973;
Kamminga and Wiersma 1981; Evans et al. 1988; Silber 1990; Amundin
1991a,b). Cranford (1992a, p.123) proposed that two terms should be used
to describe these odontocete pulse waveform variants. The term oligocyclic
describes signals containing a few cycles (often with a broad bandwidth,
bimodal, high-frequency distribution) and the term polycyclic refers to
pulses containing several cycles (with a narrowband, unimodal, highfrequency distribution).
These two signal types can also be distinguished based upon their
time/bandwidth product (TBP), as shown by Wiersma (1982, 1988). He
showed that the pulsatile signals of many odontocetes approach the theoretical curve for a minimum time/bandwidth product, and furthermore, that
these two signal types cluster near the curve in two distinct locations. It is
important to note, as pointed out by Wiersma (1988), that the tendency of
these signals toward a minimum time/bandwidth product makes them ideal
for conveying the greatest amount of information when the noise spectrum
is unknown.
Since form and function are inextricably linked, it should not be surprising that the two signal types are also associated with differences in anatomic
structure (Cranford 1992a). If we consider the relationship between the distribution of these waveform types and the anatomic configuration of the
sound source anatomy across an array of odontocete species, an interesting
relationship emerges. Apparently, species with a bilaterally symmetrical
configuration for the two sound generation complexes produce polycyclic
waveforms with a unimodal frequency distribution. However, species that
possess sound generation complexes that are bilaterally asymmetric appar1 The pressure waveforms for these directional pulsed signals can exhibit significant
distortion if they are recorded outside the axis of the primary acoustic beam. In
order to inject some consistency into the distinction of waveform types and this
discussion of them, it is useful to consider only those signals recorded within the
primary beam axis.
115
same location with homologous structures in different species (Cranford
1992a).1 Exploration of these discrete acoustic differences is correlated with
anatomic differences that suggest distinctions in the physiologic mechanism. Typically, the pulses we associate with echolocation behavior in a few
well-studied dolphin species contain only a few pressure cycles (three to
seven) in the time domain. These "typical" signals have relatively broad
bandwidths per pulse (30 to 60kHz, at -3dB; see Table 7.2 on page 134 in
Au 1993) and may have a bimodal distribution in the frequency spectrum
(Au et al. 1995; Sigurdson 1997a,b).
By contrast, the pulses produced by porpoises (Phocoenidae) and some
small delphinids, such as Cephalorhynchus commersonii (Commerson's
dolphin), contain many cycles (8 to 20) per pulse and have a narrow
bandwidth character «25 kHz at -3 dB) with a unimodal distribution in
the frequency spectrum (Dubrovskiy et al. 1971; M0hl and Andersen 1973;
Kamminga and Wiersma 1981; Evans et al. 1988; Silber 1990; Amundin
1991a,b). Cranford (1992a, p.123) proposed that two terms should be used
to describe these odontocete pulse waveform variants. The term oligocyclic
describes signals containing a few cycles (often with a broad bandwidth,
bimodal, high-frequency distribution) and the term polycyclic refers to
pulses containing several cycles (with a narrowband, unimodal, highfrequency distribution).
These two signal types can also be distinguished based upon their
time/bandwidth product (TBP), as shown by Wiersma (1982, 1988). He
showed that the pulsatile signals of many odontocetes approach the theoretical curve for a minimum time/bandwidth product, and furthermore, that
these two signal types cluster near the curve in two distinct locations. It is
important to note, as pointed out by Wiersma (1988), that the tendency of
these signals toward a minimum time/bandwidth product makes them ideal
for conveying the greatest amount of information when the noise spectrum
is unknown.
Since form and function are inextricably linked, it should not be surprising that the two signal types are also associated with differences in anatomic
structure (Cranford 1992a). If we consider the relationship between the distribution of these waveform types and the anatomic configuration of the
sound source anatomy across an array of odontocete species, an interesting
relationship emerges. Apparently, species with a bilaterally symmetrical
configuration for the two sound generation complexes produce polycyclic
waveforms with a unimodal frequency distribution. However, species that
possess sound generation complexes that are bilaterally asymmetric appar1 The pressure waveforms for these directional pulsed signals can exhibit significant
distortion if they are recorded outside the axis of the primary acoustic beam. In
order to inject some consistency into the distinction of waveform types and this
discussion of them, it is useful to consider only those signals recorded within the
primary beam axis.