Introduction & Objectives

The Navy made a commitment to a federal regulator (National Marine Fisheries Service) for monitoring within the Mariana Islands. In order to improve the Navy's analyses of impacts to marine species from training and testing activities, the Navy saw a need for improved density and abundance estimates of beaked whales which are known to be sensitive to naval sonar. 

Passive acoustic monitoring (PAM) was carried out in the Mariana Islands Training and Testing (MITT) study area in order to determine 1) the occurrence of beaked whales in the region, 2) the estimated abundance and density of the detected beaked whales, and 3) the presence and approximate direction of arrivals of any sonar detections in the PAM data. 

Technical Approach

Six Rockhopper passive acoustic recording units were deployed off the west coast of Guam in the Mariana Islands Training and Testing (MITT) study area to monitor beaked whales and mid-frequency active sonar (MFAS). Five units were arranged in an array with a 1-kilometer (km) baseline in the northern part of the survey area to determine a detection probability function through spatial capture-recapture analysis, which is crucial for estimating density. An additional sixth unit was deployed about 15 km south of the array to investigate spatiotemporal variability in beaked whale encounters within the survey area. The deployment configuration was then reversed in a following survey to study spatial variability. 

After deployment, each receiver location was surveyed to determine the location within ±3 meters. Acoustic pingers were attached to each hydrophone mooring for relative synchronization. Acoustic data was continuously recorded at a rate of 197 kHz with 24-bit resolution. 

Detection and classification of beaked whale acoustic encounters was conducted using PAMGuard software with customized click classifiers. Occurrence patterns for each species and location were generated using a custom R script. A regression analysis was employed to examine the spatial and temporal patterns of encounters at the two survey locations. The primary method for estimating beaked whale detection functions in this analysis was acoustic SCR, as described by Stevenson et al. (2015). A detection function was fitted for each species, where sample size allowed. Further, if additional auxiliary information was available for a given species, then beaked whale density was estimated around the array.

The Navy sonar analysis was performed with a highly efficient detector developed by the Naval Information Warfare Center and the Naval Undersea Warfare Center under the Living Marine Resources Program, run on the Raven-X platform (Dugan et al., 2015).

Progress & Results

Part 1:

Acoustic data was collected from 26 May 2022 to 22 November 2022. An additional ~three months of data (7 December 2022 through 14 March 2023) was collected at Site S02.

A total of 132 beaked whale encounters were registered across the two sites. Goose-beaked whales comprised 122 encounters and occurred almost exclusively at night. Using regression modeling, a comparison of beaked whale daily encounter rates at the two sites was performed. Results showed a higher number of detections in the first half of the deployment and variability in the number of detections between the single site and the array instruments.  

The density of goose-beaked whales was estimated using spatial capture-recapture (SCR). Average densities ranged from 0.83 animals per 1,000 square kilometers (km²) (coefficient of variation [CV]: 9.62%) in September 2022 to 6.41 animals per 1,000 km² (CV: 7.30%) in June 2022. Distance sampling was also investigated for goose-beaked whales through acoustic localization, though the method could not be reliably applied due to the low sample size and violation of key modeling assumptions.

A total of 16 MFAS encounters were recorded, 9 of which were detected at both sites. Six MFAS encounters were only registered at the southern (single-unit) site and one at the array site. Most of these encounters were characterized by a low signal-to-noise ratio (SNR), suggesting that these signals were likely emitted in deeper water to the south and southeast of Guam. The signals comprising the nine encounters detected at both sites occurred at distances too far from the units to obtain localizations with reasonable accuracy. For this reason, only direction-of-arrival (DoA) was calculated. 

In contrast to the beaked whale detections, most MFAS detections occurred during the day. No beaked whale detections occurred in close temporal proximity to the MFAS detections. However, it should be noted that the detection ranges for both signal types differ vastly.

Part 2: 

Acoustic data were collected from March 7, 2024, through September 4, 2024. A total of 192 beaked whale encounters were registered across the two sites (up from 132 in 2022). Goose-beaked whales comprised 115 encounters (down from 122 in 2022) and, as previously observed, occurred almost exclusively at night. Compared to 2022, there was a significant increase in Cross Seamount beaked whale (BWC) detections. A total of 70 encounters were registered across the two sites, with detections predominantly occurring during daylight hours, indicating niche partitioning with goose-beaked whales.

The density of goose-beaked whales was estimated by applying an SCR framework. Average densities ranged from 1.48 animals per 1,000 square kilometers (km²) (coefficient of variation [CV]: 8.99 percent) in August 2024 to 11.64 animals per 1,000 km² (CV: 7.01 percent) in March 2024. These numbers fall within the same range as those reported for the 2022 survey. The monthly estimates indicated higher densities in spring, decreasing over summer into fall. This pattern was also observed in 2022, suggesting some possible consistency between years, though more data are needed to establish a longer-term trend.

While a detection probability function was derived for BWC, densities could not be estimated due to a lack of auxiliary information (e.g., group size).

Sonar activity was low, with only two brief (3–4 hours) MFAS encounters observed. The direction-of-arrival analysis indicated that the signals came from due west of the deployment site. There were no beaked whale detections close in time to the MFAS events. However, it is important to note that detection ranges for both signal types differ greatly.

Overall, this study demonstrated that an array configuration with a 1 km baseline is effective for SCR beaked whale density estimation and should be maintained for similar tasks in future projects. Lastly, minor adjustments to the analysis code during this effort resulted in updates to the deployment 1 results. The revised beaked whale density estimates are presented in Appendix C of the 2025 report.