Figure 1.
Representative clicks from a) Canadian harbour porpoise, b) Danish harbour porpoise and c) Dall’s porpoise.
(Fast Fourier transform size of 512, spectrum interpolated with a factor 10, sampling rate 500 kHz, rectangular window). Note that the scale of the Y-axis in the first panel varies due to differences in received level.
Table 1.
Echolocation click source parameters of on-axis clicks from Danish and Canadian harbour porpoises (Phocoena phocoena) and Canadian Dall’s porpoise (Phocoenoides dalli) recorded with a six element hydrophone array.
Figure 2.
Centroid frequency, bandwidth (rms) and duration were used to separate BC harbour porpoises, Danish harbour porpoises and Dall’s porpoises. All parameters were significantly different across populations. The three species could be separated 84% correctly.
Figure 3.
Acoustic species discrimination.
Dall’s (circles) and BC-harbour porpoises (triangles) can be separated by means of differences in centroid frequency using a criterion of 139 kHz in a Monte Carlo simulation. The clicks were first filtered with the harbour porpoise’ audiogram (see text) to simulate porpoise reception. The dashed line indicates 90% correctly classified clicks. Such differences may also be useful in passive acoustic monitoring, provided there is fine-scale frequency resolution in the PAM dataloggers. The percentage correct (y-axis) for each click pair is the mean of ten rounds of randomly drawing 100 click pairs consisting of N clicks per pair (x-axis), and the values are shown with the standard error of the mean. The clicks included are one on-axis click for each five off-axis clicks.
Table 2.
Classification matrix (cases in rows, categories classified into columns) of the canonical discriminant analysis for the three porpoise groups.
Figure 4.
Vertical transmission beam patterns of Canadian harbour porpoise, Danish harbour porpoise and Dall’s porpoise.
A) Canadian harbour porpoise, b) Danish harbour porpoise and c) Dall’s porpoise. The points are field data. On-axis clicks recorded within 20 m from the array (5 BC harbour porpoise, 19 Danish harbour porpoise, 15 Dall’s), each with the five off-axis versions recorded on the other hydrophones simultaneously. 0–90 are degrees off-axis re on-axis at 0°. 0 to −30 is dB re on-axis source level.
Figure 5.
Source levels of Canadian harbour porpoises, Dall’s porpoise and Danish harbour porpoises.
A) Source level plotted against range to array with linear regressions for Canadian harbour porpoise (BC-HP, grey circles), Danish harbour porpoise (DK-HP, black squares), Dall’s porpoise (Dall, white triangles). B) Mean source level and standard deviation per Inter-Click-Interval (ICI) band for Danish harbour porpoises (black), Canadian harbour porpoise (grey) and Dall’s porpoise (white). Danish harbour porpoises use clicks of significantly higher source level regardless of range or ICI band than the two other porpoise groups.
Figure 6.
Effects of absorption and masking noise on detection range of NBHF and broadband dolphin style clicks.
Two effects are modeled; 1) effects of varying centroid frequency (Solid lines) and 2) effects of increasing bandwidth three times (Broken lines). The effect is calculated as summed costs/benefit in dB in relation to the echo-detection- ratio (ENR) at detection threshold (DT) of a NBHF click (ΔENR) (y axis) with bandwidth of 15 kHz as a function of target range (x-axis). Thus at 0 dB the NBHF echo is just detectable for a given range, source level and target. A positive ΔENR value means that the animal would need to increase source level by that dB difference to detect that same target. If the ΔENR is below zero it means a better echo to noise ratio by that dB difference than for the NBHF signal. The figure assumes fixed source energy flux density level and target strength (TS). Masking noise levels (NL) are calculated from fig.7.5 in [2] for sea state 3 in deep water assuming a fixed bandwidth of 15 kHz. Absorption is calculated by equations given by [36] for relevant centroid frequencies (32.5, 65, 130 and 200 kHz) at 14°C and salinity of 33o/oo. Solid lines show detection ranges for a NBHF type click with varying centroid frequency. Broken lines of colour x mimics the effect of switching to a dolphin type click with a bandwidth of 45 kHz (3×NBHF bandwidth) with the same centroid frequency and source energy flux density as solid colour x. The figure shows that NBHF click yields longer detection ranges than dolphin type clicks out to ranges of about 200 m, assuming clicks of equal source energy flux levels. This is primarily caused by the lower detection threshold effects of a narrow bandwidth signal. (See full explanation and assumptions of the model in the methods section).