Long-term and large-scale spatiotemporal patterns of soundscape in a tropical habitat of the Indo-Pacific humpback dolphin (Sousa chinensis)

Abstract

Little is known about the characteristics of ambient sound in shallow waters southwest of Hainan Island, China, a tropical habitat of the Indo-Pacific humpback dolphin. The spatiotemporal patterns of soundscape in this area were thus studied and described here. Acoustic data collected from February 2018 to February 2019 at ten monitoring sites, spanning ~200 km of the coastline, were analyzed. The ambient sound characteristics in the investigated area showed significant spatiotemporal variations. Sound levels centered at 0.5 and 1 kHz were higher during dusk and night than other times of the day at all monitoring sites except for one. Higher sound levels at frequencies above 8 kHz were documented during autumn and winter at all sites except for three of them. Biological and anthropogenic sound sources including soniferous fishes, snapping shrimps, dolphins, ships, pile-driving activities, and explosions were identified during spectrogram analyses of a subsample of the dataset. The shipping noise was frequently detected throughout the monitoring sites. Spatiotemporal variations of the soundscape in the investigated waters provided baseline information on the local marine environment, which will be beneficial to the protection of the vulnerable Indo-Pacific humpback dolphin population recently discovered in the investigated waters.

Fig 1

Map of the study area and monitoring sites (A) and picture of the acoustic platform (B). Monitoring sites are marked with ○, and different colors means different types of seabed. The major ports and other ports are marked with ☆. The SoundTrap recorder was fixed on a stainless steel column at the center of each platform.

Fig 2. Spatial variation of broadband sound pressure levels (broadband SPLs) at the PAM sites.

The differences of broadband SPLs from 20 Hz to 144 kHz among ten monitoring sites (Median; Whisker: 25^th–75^th).

Fig 3. Spatial variation of octave band sound pressure levels (octave band SPLs) at different frequencies at the PAM sites.

The differences of eleven octave band SPLs among ten monitoring sites (Median; Whisker: 25^th–75^th).

Fig 4. Comparison of broadband sound pressure levels (broadband SPLs) among different seabed environments.

The three categories were muddy, sandy and rocky sea floors. Differences are marked with *** for p-level < 0.001 (Mann-Whitney U test).

Fig 5. Comparison of octave band sound pressure levels (octave band SPLs) among different seabed environments.

The three categories were muddy, sandy and rocky sea floors. Differences are marked with *** for p-level < 0.001 (Kruskal-Wallis test).

Fig 6. Daily trends of broadband sound pressure levels (broadband SPLs) during one year at the PAM sites.

Differences among dawn, day, dusk and night are marked with *** for p-level < 0.001 (Kruskal-Wallis test).

Fig 7. Daily trends of octave band sound pressure levels (octave band SPLs) for different frequency bands during one year at the PAM sites.

Differences among dawn, day, dusk and night are marked with *** for p-level < 0.001; ** for p-level < 0.005; * for p-level < 0.05 (Kruskal-Wallis test).

Fig 8. Seasonal trends of octave band sound pressure levels (octave band SPLs) at the PAM sites.

Differences among spring (March, April and May), summer (June, July, and August), autumn (September, October and November) and winter (December, January, and February) are marked with *** for p-level < 0.001; ** for p-level < 0.005; * for p-level < 0.05 (Kruskal-Wallis test).

Fig 9. Spectrograms of the selected biological sounds.

Spectrograms of “clicks” produced by IPHD (A), the sounds produced by snapping shrimp (B) and six different fish call types (C–H). (FFT length = 144000 points, Hanning window and 50% overlap, the color-bar shows the power spectral density (dB re 1 μPa²Hz^-1)).

Fig 10. Spectrograms of the selected anthropogenic sounds.

Spectrograms of ship noise (A-B), the sound from explosion used for illegal fishing activity (C) and the sound of pile-driving activity (D). (FFT length = 144000 points, Hanning window and 50% overlap, the color-bar shows the power spectral density (dB re 1 μPa² Hz^-1)).

Fig 11. The percentage of files with fish calls in the time of day for the PAM sites.

For each site, the percentage of files that contained fish calls were documented at different time of the day by analyzing the spectrogram of three-day sound recordings during new moon period.

Fig 12. Percentage of files with ship noise.

For each site, number of files with ship noise were documented by examining the spectrogram of three-day recordings during new moon period.