Table of Contents

• Coverage
• Dataset Description
  • Methods & Sampling
  • Data Processing Description
  • BCO-DMO Curation Notes
• Parameters
• Instruments
• Project Information
• Funding

Updates are planned for this dataset to extend the time range into 2027 as part of this project.

We used Loggerhead Instruments SNAP systems and Cyclops to perform these recording experiments. Recorders were placed on a concrete mooring and strapped to a vertical PVC pipe in an upright orientation (SNAP), or positioned horizontally (Cyclops). All moorings were marked with surface flotation for initial deployments (10/11/22) but were later abandoned due to concerns regarding ambient noise generation. Recordings (.wav) were continuous, with each file set to record for 300 sec at 44.1 kHz and +2.0 DB on the gain setting. Times were recorded in local (Atlantic Daylight Time) to facilitate comparisons with other sampling events occurring on site.

SNAP recorders were initially deployed at 4 shallow, sandy spots on 10/11 (Trinity Church, Cockroach, Major's Bay, and Tucker's Bay) as these were locations where high benthic invertebrate abundances and/or foraging effects were previously recorded. These were all moved to Trunk Island and Shark Hole on 10/13 for attenuation experiments where bivalves were fractured at various (4) distances from a source. Later that same day, the four SNAP systems were redeployed across 4 new sites (Shark Hole, Middle Mount, Easter, and Crow Island) for additional ambient monitoring. Loggerhead Cyclops systems recorded audio in similar ways but only for short durations due to battery limitations. The Cyclops system was deployed at Trunk Island on 10/11, retrieved from Trunk Island on 10/14, and redeployed at the Bermuda Aquarium, Museum, and Zoo (BAMZ) on 10/17 for another ~36 h deployment. All systems were retrieved on 10/18 and memory cards downloaded for permanent storage. 

The following Loggerhead Instruments systems were used during this sampling event (with serial numbers):

* SNAP 1 = 9040036131081365
* SNAP 2 = 9040035981409044
* SNAP 3 = 9040036052785299
* SNAP 4 = 9040036452392085
* Cyclops = 0423322906173210

Data are in raw (.wav) format.   The .wav files are attached to this dataset in .zip file packages.

Data (.wav) file names represent the following fields (constant fields are in bold and surrounded by italics):

YYYYMMDD"T"HHMMSS"_"16-digit Serial number"_"GainSetting(dB)

Where Y = year, M = month, D = day, H = hour, M = month, S = second.

Note: For Cyclops audio the 16 digit serial and gain setting are not incorporated into the audio filenames, which are otherwise in the same format.

Curation Actions Performed on Data

Version 1:
* Merged .wav inventory file with deployment information
* Added .wav files to dataset page in folders per serial number
* Converted datetimes to ISO format
* Zip packages made for .wav files per sensor

Processing pipeline for primary data file 964701_v1_wavefiles.csv:

- Loaded 2022_SNAP_Deployments_withIDs.csv, deployment records with fields including Deploy_ID, Equipment_Type, Site_Name, Latitude (DD), Longitude (DD), DeployDateTime (Local), RetrieveDateTime(Local)
- Loaded sheet 2 of ajemian_file_manifest.xlsx, containing acoustic recording file manifest data
- Added computed column file_name_real copying value of file_name column on ajemian_file_manifest-2
- Applied find/replace on file_name in ajemian_file_manifest-2 to append serial number "0423322906173210" to file names matching a date-only pattern (not already containing an underscore-separated suffix), standardizing Cyclops hydrophone file names
- Split file_name column into date, serial_number, wav columns using underscore delimiter
- Split date column further into date and time columns using "T" delimiter
- Split DeployDateTime (Local) column in ajemian_file_manifest-1 into deploy_date and deploy_time columns using space delimiter
- Split Deploy_ID column in 2022_snap_deployments_withids into date, location, equipment, number columns using underscore delimiter, preserving field metadata
- Added computed serial_number column to 2022_snap_deployments_withids by matching equipment type and number to known device serial numbers (SNAP 1-4 and Cyclops), to enable joining deployment records to file manifest records by serial number and date
- Joined 2022_snap_deployments_withids (source, deleted after join) into ajemian_file_manifest-2 (target) on matching serial_number and date columns, using half-outer join mode, pulling in DeployDateTime (Local), Deploy_ID, Equipment_Number, Equipment_Type, Latitude (DD), Longitude (DD), RetrieveDateTime(Local), Site_Name, and number columns
- Reordered columns in ajemian_file_manifest-2
- Deleted columns file_name, "BCODMO file path brooks", date, wav, time, Equipment_Number from ajemian_file_manifest-2
- Converted DeployDateTime (Local) from string (format %m/%d/%Y %H:%M, Atlantic/Bermuda timezone) to datetime output (format %Y-%m-%dT%H:%M, Atlantic/Bermuda timezone)
- Converted RetrieveDateTime(Local) from string (format %m/%d/%Y %H:%M, Atlantic/Bermuda timezone) to datetime output (format %Y-%m-%dT%H:%M, Atlantic/Bermuda timezone)
- Set column types on ajemian_file_manifest-2: DeployDateTime (Local) and RetrieveDateTime(Local) as datetime, Deploy_ID/Equipment_Type/Site_Name/file_name_real as string, Latitude (DD)/Longitude (DD) as number, number and serial_number as integer
- Renamed columns: Latitude (DD) to Latitude, Longitude (DD) to Longitude, DeployDateTime (Local) to DeployDateTime_Local, RetrieveDateTime(Local) to RetrieveDateTime_Local, number to Equipment_Number, serial_number to Serial_Number, file_name_real to File_Name
- Renamed resource ajemian_file_manifest-2 to 964701_v1_wavefiles
- Saved as final table: 964701_v1_wavefiles.csv

Version 2: Summary of Changes
- Dataset version 2 started to receive extended dataset coverage to 2022-2023. But as of 2026-09-04 this page contains 2022 data and metadata.
- Updated previous zipped files drafted in version 1 of this datset with current set as provided in a Globus shared folder Ajemian_UW_Crunchtime/PAM/BDA 2022 folder. Files were indexed and sent to the data submitter for review.
- Added a Deployments table (from deployments.xlsx, as submitted, exported to deployments.csv), this table contains 2022 records only as of 2026-09-06.
- Renamed 964701_v1_wavefiles.csv to 964701_v2_wavefiles.csv to reflect the version update. No changes were made to its contents, since no additional metadata was submitted for the 2022 files.
- four per-sensor voltage/firmware log files were attached as supplemental files as submitted (20221011_S1_LOG.CSV through S4_LOG.CSV).

Curation Actions Performed on Metadata

- Standard metadata entry and formatting of the metadata as submitted.

Known Issues Potentially Impacting Reuse

- Data restriction 2028-03-31 may not be in line with NSF 2-year release policy; no confirmation was received if this was discussed with the NSF program manager.

- deployments.csv, as submitted, contains only 2022 records (all October 2022); 2023 deployment metadata has not been submitted, despite the dataset title/description covering both years.
- 964701_v2_wavefiles.csv, as submitted, has not been updated to reflect the six newly assembled 2022 ZIP packages.

- 20221011_S4_LOG.CSV contains residual entries dated 2020-12-18/19, preceding the October 2022 entries.
- 20221011_S2_LOG.CSV contains a duplicate header row and an anomalous early file (20221011T081000) about four hours before other sensors' first files that day, possibly a test recording rather than a deployment recording.

NSF Award Abstract:
Shellfish (mollusks, crustaceans, etc.) are facing unprecedented pressures under global climate change, which is threatening the variety of ecosystem services these animals provide to coastal communities. While much research has been dedicated to understanding how changing ocean conditions can influence shellfish development, far less has explored the potential impacts from increasing populations of large, shell-crushing predators (i.e., rays, turtles, etc.) that are experiencing poleward expansions of their ranges. This knowledge gap is likely due to the challenges of working with these mobile species, which require novel technology to track their dynamic distribution and thus foraging effects on shellfish communities. This project will build fundamental knowledge on marine habitats susceptible to predation from large mobile predators in order to ensure a sustainable future for shellfish species. Further, the work will provide guidance to costly shellfish restoration programs that are otherwise “flying blind” with respect to predation risk. The project will have local, regional, and global educational dimensions. Firstly, this project will strengthen FAU’s graduate programs by supporting a graduate student and providing a platform for the PI to develop a new graduate course, which will be offered and evaluated twice throughout the award period. Additionally, numerous undergraduate summer interns and middle-high school students will be recruited to interact with the PI via immersive, hands-on field excursions. Lastly, the fascination of the general public and students with these charismatic animals and the project’s tangible technological components will facilitate developing an interactive “Audio Waves” exhibit at a local outreach center, which will be evaluated several times during the project and slated for permanent display.

Our scientific understanding of the ecological role of large mobile durophages (i.e., shell-crushing predators) is limited due to challenges presented by the elusive nature of these species. These shortcomings hinder our scientific understanding of their role in benthic community dynamics. Filling such knowledge gaps requires novel approaches that can detect and classify predator-prey interactions in situ. Using multiple large predator models (rays, sea turtles, fish, and crabs), the project will: 1) capture and characterize predator feeding (shell-crushing) sounds and shell fragmentation patterns, 2) understand in situ detection constraints of the predation signal within the context of natural underwater noise using simulations, and 3) quantify the distribution of predator foraging impacts across two model seascapes in Bermuda and Florida via integration of habitat- and individual-based (animal tags) passive acoustics. Detection and classification (by both predator and prey) will be completed using novel application of machine-learning techniques, which will be used to automate predation event extraction from extensive data archives. Recording equipment will be strategically distributed across seascapes to permit a multi-scale understanding of durophagy and testing of theoretical models of predation (e.g., optimal/central place foraging). Long-term monitoring will also provide an opportunity to assess the role of environmental/oceanographic variables in driving these interactions. Consequently, this work will fill a large knowledge gap in the dynamics of marine food webs.