Soundscape measurements
All the recordings used during the experiment were recorded in two regions of the North Sea: the Southern Bight near the Belgian coast and in the Dutch Wadden Sea (see Figure 1). Details of the data collection are explained in the Supporting Information (Appendix S1). The recorded data were manually scrolled through to select suitable files for the off-reef and the vessel treatments. Only data from spring and summer were considered, in order to correspond to the sounds from the Wadden Sea (reef and off-reef). Selected data are listed in the Supporting Information Appendix S1, Table S2. In total, 3 recordings of reefs from 2 different locations, 4 vessel recordings with several boats on each recording from 4 different locations, and 4 off-reef recordings from 3 different locations were used to represent our treatments.
Sound files
The collected sounds were scanned to select appropriate sound snippets. These segments were selected to be representative of each treatment. For example, reef sounds were only selected when they contained no apparent outside influences (e.g. vessel sounds). For the vessel sounds, a fair variability of sounds was selected, from short sounds of distant vessels to longer continuous sounds from vessels operating close by, with no other audible background sounds. The selected segments were then combined to create one 1 h file per treatment and day. In some cases, the selection led to files shorter than 1 h, so the segments were repeated and combined by applying crossfading with Audacity35 to create a 1 h file. When enough recordings were available for 1 h or more, segments were not repeated.
Throughout the experiment, the treatment groups were consistent but the sound file differed in each replicate. For reef treatment, sounds used were recorded from the same location in Texel, NL but sound files used during each day of the experiment were selected from different recording dates (see Table S2). For the off-reef sounds two sound files were used recorded from Texel, NL and two sound files were used recorded from non-reef areas in the Southern Bight off the coast of Belgium (see Table S2). All vessel sounds were recorded from locations in the Southern Bight (see Table S2). Treatments where vessel sounds and reef sounds were played together were created artificially overlaying the reef sound file and the vessel sound file. The use of multiple sound files of the same treatment was used to strengthen confidence that the sounds were representative for the overall soundscape and not for a single event.
The files used were acquired with different instruments and at different locations. To deal with the difference in sampling rate and minimum recording frequency, all the files were filtered using a butterworth bandpass filter (N=4) between 20 Hz and 12 kHz. After the filtering, all the files were downsampled or upsampled to 48 ksps to match the playback requirements. Information about the instruments used for recording can be found in Table S1.
Broodstock and Larvae Culture
Ten mature adult oysters (five females and five males) were purchased from the Guernsey Sea Farms Ltd (Guernsey, UK) and used to produce larvae. Eggs were fertilized by gonad stripping following FAO guidelines36. Fertilized eggs were kept undisturbed in flat bottom tanks for 48 hours at 22 °C at a density of ten eggs per ml of filtered seawater (FSW). All seawater used in this experiment was filtered at 0.1 µm and passed through UV light. After 48 hours larvae were sieved over 70 µl nylon mesh, rinsed, and transferred to rearing tanks with FSW. Tanks were aerated and kept at 22 °C for the entire duration of larvae rearing. Every two days larvae were sieved over mesh corresponding to the average size of the larvae and the water in the tanks was changed. Larvae were fed a mixture of fresh microalgae mixture consisting of Chaetoceros muelleri, and Isochrysis galbana (clone T-ISO). For the first 4 days larvae were fed at 40,000 cells/ml water using only I. galbana (clone T-ISO). Days 5-12 larvae were fed C. muelleri, and I. galbana (clone T-ISO) at 100,000 cells/ml at a volume ratio of 1:1. Days 13+ larvae were fed C. muelleri, and I. galbana (clone T-ISO) at 100,000 cells/ml at a volume ratio of 3:1. Larvae entered their pediveliger stage and became competent to settle at 29 days. Larvae were determined for competence when they had a prominently displayed eyespot and larval foot and were sized at 320-350 μm in diameter.
Settlement Experiment Design
The experiment consisted of five sound treatments: oyster reef sounds, vessel noise, reef sounds with added vessel noise, off-reef sounds, and a no-sound control. Larvae were exposed to each of these treatments in parallel, with trials that lasted 24 hours, these trials were replicated four times over four days. On each day, larvae were assessed for settlement and then discarded. At the start of each experiment day, new pediveliger stage larvae were used. Sound treatments took place in separate tanks. In each tank, 5 jars each containing 10 larvae, were used as subreplicates (see Figure 2).
Tank set-up
Five 100L tanks were used (49x65x50.5 cm), separated 20 cm from each other on a rack. Each tank sat upon a 4 cm layer of polystyrene to isolate it from the rack and an additional layer of acoustically absorbent foam (25mm thick) between the polystyrene and the tank bottom. The acoustic foam was also placed at the tank sides. Four Lubell UW30 Underwater Speakers with custom-made amplifiers, battery-powered in order to avoid 50 Hz noise, were used. Each speaker was connected to one TASCAM playback device which played on repeat a 1 to 2 h file. No speaker was placed in the no-treatment control (see Figure 2). The speakers were hung in the middle of the tank with ropes so they would not touch the tank walls. Larvae were placed inside 100 ml polystyrene jars and these containers were fixed in the same position in the tank for every day of the experiment.
Settlement Assays
Oyster larvae were reared in a laboratory scale hatchery in the same facilities as where the experiments were conducted. A detailed account on the larviculture can be found in the Supporting Information.
On each day, 10 larvae were gently pipetted randomly into each of the five 100 ml containers per tank and filled with filtered seawater (FSW) and 0.2 grams of oyster shells which could act as a settlement substrate. To get a consistent shell topography, shells were crushed using a hammer and crushed shells were sieved between 1.0 mm and 0.5 mm metal sieve. For each treatment tank, 5 individual containers were used. As all treatments were repeated over 4 consecutive days, 20 jars were used per treatment in total. All trials were conducted in a dark environment at 20 (±1) °C in a climate-controlled room.
To avoid any air in cups containing larvae, larvae were placed in the cups and the lid was fixed while the cup was fully submerged in FSW. This step was necessary to prevent any distortion of the sounds due to reflection from air bubbles. All FSW used in the experiment had added microalgae Chaetoceros muelleri, and Isochrysis galbana (clone T-ISO) at 100,000 cells/ml at a volume ratio of 3:1. In a previous study, M. gigas larvae increased swimming when exposed to reef sounds, but only if larvae were fed21, thus microalgae were added to our larvae containers. Microalgae were added at the same concentration as used in larvae rearing tanks and food levels were not limiting for the duration of the experiment.
On top of each tank, cups were attached to a wooden pole sitting horizontally across the tank. Each larvae jar was attached so that it was in a fixed position for the duration of the experiment, the position of the jar was noted so that the effect from placement in the tank could be ruled out. The wooden pole was isolated from the tank walls with polystyrene to avoid vibration propagation. One of the cups was located directly above the speaker and the other 4 cups were at the same distance from the center of the speaker (see Figure 2).
After 24 hours of exposure, larvae metamorphosis was checked using a dissecting microscope and the number of larvae that had cemented themselves to the substrates were counted. Metamorphosis was confirmed by gently blowing water over the larvae with a pipette to ensure that larvae were fixed to the substrate.
Playback and sound characteristics
For each treatment, a playback volume was chosen so the exposure power spectral density (PSD) would match the sound levels specified in literature as typical of reefs (at 1 m from the seafloor) and off-reefs (at 2 km from the reef)11,37. Details of the process done to achieve this are specified in Supporting Information.
To quantify the exposure sound level and the acoustic characteristics of each playback, each treatment was recorded using the chosen playback volume for 1 h (experiment files) at 48 ksps. When recording these 1 h files, all four different sound treatments of that batch were on to record possible acoustic crosstalk from the other treatments. These 1 h recordings were used to compute the exposure acoustic metrics for each treatment. The no-sound treatment was also recorded while all the other treatments were on. Furthermore, the room noise was also recorded using the same protocol when no speaker was active.
For each treatment, several acoustic features were computed for both the 1 h experiment files and the 1 h field files. Acoustic Complexity Index (ACI), Acoustic Evenness Index (AEI) and Acoustic Diversity Index (ADI) were computed using the maad python package38, and the Power Spectrum Density (PSD) was computed using the scipy python package39. The average PSD was computed for three different bands by averaging the spectrum density of all the frequency bins included in the specified frequency band. Parameters used to compute each of the features are summarized in Table S3.
Both ACI and ADI are proxies to quantify acoustic complexity (the higher the number, the more complex), while low values of AEI represent an even sound and higher values represent more uneven sounds. This is not correlated with the ecological concept of evenness, as acoustic evenness refers to an even distribution of sound energy in different frequency bands, and this can be achieved due to a high biodiversity vocalizing at the same time covering all the frequency bands or by constant broadband sounds such as some anthropogenic sounds40,41.
Statistical analyses
A generalized linear mixed-effect model was created using the glmer function of the lmer package42 in R version 4.1.3 (2022-03-10) (R Core Team, 2021). As the binary response variable was binary (settled vs. not settled) we fitted a Bernoulli distribution using a logit link function. The assumptions of the model were met. N value was between 175 and 180 for each treatment. The sound treatment was the only main effect variable. Individual speaker-playback, the tank used, the position of the ‘larvae container’ within the tank, and the date of the trial (as a factor) were added as main effects to the generalized linear models to determine whether these confounders had any effect on settlement. As there was no significant main effect of speaker, tank, or cup position, it was assumed that they did not have an effect on the experiment outcome and were not included in the final model. The effect of date of the experiment was significant and therefore included as a random effect variable in the final model. Post hoc tests were performed using the emmeans function of the lsmeans package43 to calculate the marginal means adjusting p-values for multiple comparisons with Tukey's method and the pairs function was used to display pairwise comparisons. Post hoc analyses were also conducted on models where experiment day was included as a main effect variable in order to ensure that this variable did not interfere with the treatment.