Three killer whale (Orcinus orca) ecotypes are found in the northeastern Pacific: residents, transients, and offshores. Each ecotype produces distinct, stereotypic pulsed calls that serve as acoustic indicators of presence. To investigate the spatial and temporal occurrence of each ecotype in the Gulf of Alaska, long-term acoustic recorders were intermittently deployed from 2011 to 2019 at seven sites along the continental shelf and slope, in an abyssal region, and near offshore seamounts. This dataset contains examples of killer whale calls from these recordings. The examples are grouped based on the results of ecotype discrimination using established catalogs of call repertoires for each ecotype.

Passive acoustic monitoring was conducted in the Navy’s Gulf of Alaska Temporary Maritime Activity Area (GATMAA) using High-frequency Acoustic Recording Packages (HARPs) from July 2011 to September 2019. HARPs were deployed in a seafloor mooring configuration with the hydrophones suspended at least 10 m above the seafloor, except for Abyssal where the seafloor was at 4,000 m and the hydrophone at 1,200 m. Each HARP hydrophone was calibrated in the laboratory to provide a quantitative analysis of the received sound field. Representative data loggers and hydrophones were also calibrated at the Navy’s Transducer Evaluation Center facility to verify the laboratory calibrations.

The seven sites monitored spanned the continental shelf (sites Kenai Shelf and Kodiak Shelf), the continental slope (sites Kodiak Slope and Kenai Slope), deep, offshore water (site Abyssal), and offshore seamounts (sites Quinn and Pratt). Kenai Shelf was monitored intermittently from July 2011 to September 2014, Kodiak Shelf almost continuously from June 2013 to September 2014, Kodiak Slope from April to September 2019, Kenai Slope intermittently from July 2011 to September 2019, Abyssal from April to September 2017, Quinn intermittently from June 2013 to September 2017, and Pratt almost continuously from September 2012 to September 2014.

Each HARP sampled continuously at 200 kHz except for one deployment each at Kenai Shelf and Kenai Slope, where a duty cycle was used, and one deployment each at Kenai Shelf and Quinn, which were sampled at 320 kHz. Duty-cycled deployments recorded for 10 minutes on and 2 minutes off. There were 4,545 days of recording effort overall.

Before analysis, data were decimated by a factor of 20 for more effective scanning up to 5 kHz (8 kHz for deployments that were sampled at 320 kHz). Long-term spectral averages (LTSAs) were created using a 5-s time average and a 10-Hz resolution and were scanned by an analyst for killer whale whistles and pulsed calls using Triton, custom MATLAB (MathWorks, Natick, MA) software. When a potential killer whale signal was identified in the LTSA, a 30-s long spectrogram (1000-point FFT length, 90% overlap) was examined to confirm that the calls belonged to killer whales. Calls were logged as encounters, where each encounter was bounded by at least 15 min of recording that did not contain any calls.

 Each encounter was later examined for pulsed calls which, when found, were scrutinized in a 30-s spectrogram window in order to visually and aurally identify call types that could be used to distinguish between different ecotypes. An encounter was only classified to ecotype level if calls from the encounter could be positively matched to existing call repertoires. Resident and transient call types were identified using a digitized call catalog made from recordings provided by H. Myers (North Gulf Oceanic Society, unpublished data). These call types were from previously published call catalogs and from vessel surveys with resident killer whales (North Gulf Oceanic Society, unpublished data). Resident and transient encounters were confirmed by H. Myers and D. Olsen. Offshore encounters were confirmed by J. Pilkington by matching calls to a digitized call catalog of offshore calls made from field recordings taken during encounters with photo-identified offshore killer whales (DFO Cetacean Research Program, unpublished data). Encounters that could not be identified to the ecotype level were labeled unknown. Unknown encounters were typically short in duration and contained short and faint calls that were not suitable for making ecotype identifications.

Please view the README.txt file for more information.