Fig 1.
(A) Location of study sites on Eastern Shore of Virginia (USA).
(B) 3 habitat types near and within Hillcrest Oyster Sanctuary were chosen for comparison – a natural oyster reef, restored oyster, and bare mudflat. Reprinted from Volaric et al. (2018) under a CC BY license, with permission from Inter-Research Science Publisher, original copyright 2018. Aerial image credit: USGS, USDA, The National Map: Orthoimagery.
Fig 2.
(A) Restored reef depicting observational setup which include a hydrophone attached to recorder and battery, wave gauge, and acoustic Doppler velocimeter (ADV).
(B) A natural reef and (C) mudflat were also sampled with the same instrument array. (D) Close up showing hydrophone extending out of metal tubing a fixed distance above the reef. These images depict these intertidal sites during emergent conditions, but data was only used when they were submerged. Image credits: M. Volaric.
Fig 3.
(A) Example power spectra over 1 high tide period from restored reef taken the morning of June 14, 2018.
This spectrogram is representative of all three sites, with SPL at higher frequencies showing a consistent relationship with depth, resulting in distinctive “canoe-shaped” spectrograms in which there was very little sound energy in the 1000 - 3000 Hz range at the beginning and end of each tidal period. (B) Sound power spectra from the same data set, shown at times 6 h in red, and 8.5 h in black. The blue line is a fit in which SPL, in Pa, is proportional to f−6.3. (C) A 10 s long spectrogram from the same data set from time 8.5 h, showing that the sound arises from a series of distinct clicks, not from a continuous noise source.
Fig 4.
Mean FN or SPL for each of the three frequency bands for each site.
Datapoints represent mean of 15 min averages ± SD. (A) FN < 100 Hz, (B) 150 Hz < SPLMID < 1500 Hz, (C) SPLHI > 7000 Hz.
Fig 5.
Example data from the mudflat showing the impact of a severe thunderstorm (occurring approximately at time 20.25–20.75) on (A) flow velocity, (B) significant wave height HS, and
(C) FN. Note higher FN at the beginning of the tide when velocity was high, then a spike in FN coinciding with a massive increase in Hs with the thunderstorm, followed by lower FN after the storm when velocity was lower. This example demonstrates how we were able to correlate time-averaged values of hydrodynamic parameters to the soundscape. In this example 3 min averaging was used to better demonstrate the impact of the storm, but all other results utilized 15 min averaging in order to allow sufficient time to calculate turbulence statistics. There were high amounts of rain during this period, which can also be detected at these frequencies [25].
Fig 6.
(A–C) Exponential fits of turbulent kinetic energy (TKE) vs. turbulence dissipation (
ε) and (D, E, F) linear fits of flow speed (U) vs. TKE. TKE vs. ε was similar at all three sites, but the U vs. TKE slope was significantly steeper at the reef sites than the mudflat due to higher benthic roughness. Each datapoint represents a 15 min average. Only data with significant wave height < 5 cm were used at the mudflat site to avoid possible wave artifacts.
Fig 7.
(A) Flow speed (U), (B) turbulent kinetic energy (TKE), and (C) turbulence dissipation (ε) vs. FN at the three sites.
These data were fit with exponential functions. Panel C also shows the theoretical FN vs. ε relationship predicted by Eqs. 10–12. Both TKE and ε had stronger effects on FN than U, as water column turbulence results in pressure fluctuations that are recorded by the hydrophone. An outlier point from the mudflat (asterisk) was not included. Please note for this comparison FN has been converted from dB to Pa.
Fig 8.
(A) Flow speed (U), (B) turbulent kinetic energy (TKE), and (C) turbulence dissipation (ε) vs. FN over a representative submerged high tide period at the natural reef.
Panel C also shows the theoretical FN vs. ε relationship predicted by Eqs. 10–12. Although correlations between FN and hydrodynamic parameters were relatively weak when integrating over all data (Fig 6), this effect is likely due to differences in background noise between different submerged periods. During each submerged period, which better controls for background noise levels, correlations between FN and hydrodynamics were much stronger. Please note for this comparison FN has been converted from dB to Pa.
Fig 9.
Significant wave height (Hs) vs. FN at all three sites.
Hs was significantly (p < 0.001) related to FN at all three sites, showing the impact of wave orbitals on low frequency pressure fluctuations recorded by the hydrophone. Note the different scale at the mudflat (C), as conditions were significantly wavier during measurements at this site. The two datapoints at high values of Hs in (C) represent the storm depicted in Fig 3. Data were averaged over 15 min intervals.
Fig 10.
Mean (A) FN and (B) bed shear stress (
τb) at all three sites ± SD. If FN is caused in large part by turbulent pressure fluctuations, it should be of similar magnitude as τb [54].