Figure 1.
Humpback whale singing activity and occurrence of seismic survey pulses.
The number of singers present in the first ten minutes of each hour was counted from spectrographic displays, and thus singing activity is represented by the mean minimum number of singers per hour recorded for both MARU hydrophones combined (accounting for whales heard on both MARUs simultaneously as a single singer). Seismic survey activity is overlaid as the number of hours in a given day during which pulses were detected, scored as present or absent for the same 10-minute periods in which singers were counted.
Figure 2.
Examples of humpback whale song and seismic survey pulses of varying received level (RL).
Each panel represents a spectrogram (FFT 512, same frame size, 75% overlap, Hann window) of a different period recorded on MARU 1. From top to bottom, the panels illustrate a 60-second sequence from a 10-minute sample period in which the measured pulse RL (re: 1 µPa2 in a 1 Hz bin, Peak Power) was (A) 109.8 dB, (B) 101.8 dB, and (C) 99.2 dB, representing the upper 1%, 5%, and 10% of all measured pulses, respectively. Power spectral density levels (in dB re: 1 µPa2/Hz) are shown to the right of each spectrogram, averaged across four consecutive 1-sec intervals beginning with the onset of the pulse (indicated on the spectrogram with colour coding, i.e., given the onset of pulse = 0 sec, averaged spectra are shown for 0–1 sec, 1–2 sec, etc.). It can be seen that pulses were approximately 10–11 seconds apart with most spectral energy in the 0–500 Hz bandwidth; reverberation lasting approximately 2 seconds was evident only on the loudest 1% of pulses, whereas 90% of pulses had a lower RL than illustrated in the bottom panel, with no detectable reverberation beyond 1 second.
Figure 3.
Received level (RL) of seismic survey pulses.
Plotted is the relationship of the measured Peak Power variable (in dB re: 1 µPa2 in a 1 Hz band for the frequency bin with the highest power; used in the GAMM analysis) to the Peak-to-Peak amplitude (in dB re: 1 µPa) for all measured seismic survey pulses.
Table 1.
Results for Generalized Additive Mixed Models fit to data from MARU 1 and MARU 2.
Figure 4.
Impact of seismic survey pulse RL on humpback whale singing activity for the full dataset.
Generalized Additive Mixed Models of the number of humpback whale singers with smooth terms for the dependence on Survey Day, Hour, Moon Phase and Peak Power fitted for each of the MARUs; plots show the estimated conditional dependence of humpback whale singer numbers on Peak Power for (A) MARU 1 and (B) MARU 2. The x-axis in each plot shows Peak Power, describing received level of seismic survey pulse (in dB re: 1 µPa2 in a 1 Hz frequency bin) with a rug plot (short vertical bars) indicating the Peak Power values of observations. The y-axis, with scale selected optimally for each plot, shows the contribution of the smooth of Peak Power to the fitted values of singer number. Estimates (solid lines) are shown with 95% confidence bands (dashed lines), indicating a significant downward trend in singer number with increasing pulse RL.