Skip to main content
Dryad logo

Changes in sea ice and range expansion of sperm whales in the eclipse sound region of Baffin Bay, Canada

Citation

Posdaljian, Natalie (2022), Changes in sea ice and range expansion of sperm whales in the eclipse sound region of Baffin Bay, Canada, Dryad, Dataset, https://doi.org/10.5061/dryad.c2fqz619z

Abstract

Sperm whales (Physeter macrocephalus) are a cosmopolitan species but are only found in ice-free regions of the ocean. It is unknown how their distribution might change in regions undergoing rapid loss of sea ice and ocean warming like Baffin Bay in the eastern Canadian Arctic. In 2014 and 2018, sperm whales were sighted near Eclipse Sound, Baffin Bay: the first recorded uses of this region by sperm whales. In this study, we investigate the spatiotemporal distribution of sperm whales near Eclipse Sound using visual and acoustic data. We combine several published open-source, data sets to create a map of historical sperm whale presence in the region. We use passive acoustic data from two recording sites between 2015 and 2019 to investigate more recent presence in the region. We also analyze regional trends in sea ice concentration dating back to 1901 and relate acoustic presence of sperm whales to the mean sea ice concentration near the recording sites. We found no records of sperm whale sightings near Eclipse Sound outside of the 2014/2018 observations. Our acoustic data told a different story, with sperm whales recorded yearly from 2015-2019 with presence in the late summer and fall months. Sperm whale acoustic presence increased over the 5-year study duration and was closely related to the minimum sea ice concentration each year. Sperm whales, like other cetaceans, are ecosystem sentinels, or indicators of ecosystem change. Increasing number of days with sperm whale presence in the Eclipse Sound region could indicate range expansion of sperm whales as a result of changes in sea ice. Monitoring climate change-induced range expansion in this region is important to understand how increasing presence of a top-predator might impact the Arctic food web.

Methods

Historical Sperm Whale Distribution in Baffin Bay

Sighting data was compiled from eight published datasets and studies to gain a better understanding of the historical distribution of sperm whales in Baffin Bay. Sampling effort for the sighting data was only available for two of the eight data sets. First, for the Programme Intégré de recherches sur les oiseaux pélagiques (PIROP) data set (CWS 2021), effort was calculated by plotting the boundaries of where sperm whales were not sighted during their surveys between 1970-2000 (Gjerdrum et al. 2012). Second, data points for sperm whale sightings and survey effort area from Shell’s marine mammal visual monitoring and mitigation program from 2012-2014 were extracted and replicated from Frouin-Mouy et al. (2017) using WebPlotDigitizer (Rohatgi 2017).

Although there was no sampling effort available for the remaining six data sets they were included in the analysis as opportunistic sightings. Commercial whaling records for sperm whales were accessed on the Ocean Biodiversity Information System (OBIS) from the History of Marine Animal Populations (HMAP) Data Set 04: World Whaling (Smith 2021). Opportunistic and incidental sightings of sperm whales were also retrieved from the Maritimes Region Whale Sightings Database (WSDB) from Fisheries and Oceans Canada (Whalesightings Database 2021), the Incidental Sightings of Marine Mammals data set from the Institute of Marine Research (IMR) (Hartvedt 2020), and from environmental surveys done by the Northwest Atlantic Fisheries Organization (NAFO 2014). Opportunistic sightings from citizen scientists were included from the Happywhale database (Happywhale 2021). One additional opportunistic sighting of a sperm whale in Eclipse Sound and tag data from sperm whales in Baffin Bay reported in Lefort et al. (2022) were also used.

Acoustic Data Collection

We used passive acoustic recordings from two sites in the eastern Canadian Arctic near Eclipse Sound between July 2015 and September 2019 over the span of five deployments (Fig. 1, Table 1). The temporal coverage among sites and between deployments varied as a result of field work timing to retrieve/deploy the instruments, the battery life and storage space of the instruments, and different duty cycle regimes discussed in further detail in the last Methods subsection. The two sites were approximately 23 km apart and have an approximate maximum detection range of 20 km for sperm whales based on previous studies (Tran et al. 2014). The Guys Bight (GB) instrument had one deployment at a depth of ~100 m in 2015. The depth of the instrument is not ideal for recording a deep-diving animal such as the sperm whale, but the recordings from 2015 were still included in the analysis as an opportunistic data set, in the sense that it was not necessarily acquired for this specific study. The instrument at Pond Inlet (PI) was at depths between 670 and 800 m over four deployment periods from 2016-2019 (Table 1). We used data from two types of autonomous, bottom-mounted recording devices to collect passive acoustic recordings: Song Meter SM2M (SM2M; Wildlife Acoustics Inc, Concord, MA, USA) and High-frequency Acoustic Recording Package (HARP; Wiggins and Hildebrand, 2007). The SM2M was deployed at the GB site and recorded at a sampling rate of 192 kHz. The HARPs were deployed at the PI site and collected recordings at a sampling rate of 200 kHz. These two sampling rates were chosen to detect the high-frequency echolocation clicks of marine mammals, including but not limited to, sperm whales.

Acoustic Data Analysis at the Guys Bight Recording Site

Sperm whales regularly produce high-frequency echolocation clicks (2 - 32 kHz) with a duration of 100 μs and are distinguishable from other high-frequency odontocetes (Weilgart & Whitehead 1988; Goold & Jones 1995; Møhl et al. 2000, 2003) (Fig. 2). Trained analysts (NP and CS) manually screened the acoustic recordings from the GB site for sperm whale echolocation clicks using long-term spectral averages (LTSAs; averaged over 5 s and 100 Hz bins), which provide a compressed spectrogram view allowing large time series data sets to be analyzed (Wiggins & Hildebrand 2007) (Fig. 2a). Data were manually scanned in the custom software program Triton (Wiggins & Hildebrand 2007) developed in MATLAB (MathWorks, Natick, MA). Analysts viewed 1-hour LTSA segments across a frequency range of 0 to 40 kHz to identify potential sperm whale encounters. Spectrograms of 10 s in length were used to confirm species identification (Fast Fourier transform length 2000 points, 75% overlap, bandwidth 40 kHz). A sperm whale encounter was defined as a series of clicks within a recording period of 5 minutes. The start and end times of these encounters were logged in Triton (Wiggins & Hildebrand 2007) and used for further analysis.

Since all LTSA detections were visually verified by trained analysts, we assumed there were no false positives. And given the distinguishable echolocation clicks of sperm whales and the short duration of the data, we assumed negligible missed detections because of the feasibility of listening to and viewing all spectrograms of interest.

Acoustic Data Analysis at the Pond Inlet Recording Site

Clicks at the PI site were automatically detected using a multi-step approach (Roch et al. 2011; Soldevilla et al. 2011; Solsona-Berga et al. 2020). This approach was developed for acoustic data that was sampled at 200 kHz (PI recording site), which is why manual analysis was conducted for the GB site that was sampled at 196 kHz (Table 1).

Since sperm whale clicks are characterized by multiple pulses approximately 5 ms apart (Møhl et al. 2000), clicks closer than 30 ms apart were merged. Band passing all acoustic data (5-95 kHz) minimized the effects of low-frequency noise from vessels or weather. Calculating the spectra of detected signals required 4 ms of data and a 512-point Hann window centered on the click with 50% overlap. To account for the frequency dependent instrument response of each deployment, spectra were corrected for the hydrophone transfer function.

Sperm whale echolocation clicks are similar to the impulsive signals from ship propeller cavitation, so an automated detector was used to exclude periods of ship passages in the PI data to reduce the number of false positive detections. The detector, developed by Solsona-Berga et al. (2020) identified potential ship passages from LTSAs. Average power spectral densities (APSD) were computed in 2-hour data blocks over low (1-5 kHz), medium (5-10 kHz), and high (10-50 kHz) frequency bands. Using received sound levels, transient signals such as odontocetes, ship passages, and weather, were compared in the three frequency bands. Trained acousticians (NP and CS) manually verified identified ship passages in Triton (Wiggins & Hildebrand 2007; Solsona-Berga et al. 2020). Ship passage times were removed from further analysis and considered time periods with no recording effort.

Noise produced by the instruments and clicks produced by non-sperm whale odontocetes were also removed in the PI data to reduce the number of false positive detections. A basic classifier using spectral click shape and peak frequency was implemented, taking advantage of a sperm whale click’s distinctive lower frequency spectral shape to remove clicks by delphinids and beaked whales which occur at higher frequencies (Solsona-Berga et al. 2020). The remaining acoustic encounters at site PI, containing presumed sperm whale echolocation clicks, were manually reviewed with DetEdit (Solsona-Berga et al. 2020). DetEdit provides users with interactive data visualizations that aid in efficiently annotating data, allowing deletion of false detections. Sperm whale clicks were binned into 5-minute intervals, and the number of 5-minute bins with sperm whale detections per day was considered for further analysis. Daily averages of 5-minute bins were calculated only for days with sperm whale presence since the time series was zero-inflated.

Only clicks exceeding a peak-to-peak received level (RL) of 125 dBpp re 1 µPa were included to provide a consistent detection threshold. This step of the detection process excluded clicks with low RLs. However, by binning clicks into 5-minute intervals, the chances of missing presence within a bin was low for sperm whales who click regularly.

Accounting for the Duty Cycle

Duty cycle regimes, or the process of turning an acoustic recorder on at specified intervals, are implemented to maximize the deployment duration by conserving battery power and storage space of the instrument (Wiggins & Hildebrand 2007; Au et al. 2013). Duty cycle regimes can widely vary based on the desired deployment duration, the sampling rate, and the recording instrument. Two of the five deployments in this study had a duty cycle that was adjusted for.

The GB deployment had a duty cycle with a 5 minute recording duration in a 60 minute cycle (Table 1). Because there was only one deployment at this site, we had no continuous data to quantify the impact of the duty cycle. Instead, the number of 5-minute bins with sperm whale detections per day was linearly adjusted based on the recording effort in the duty cycle.

The third deployment at the PI site also had a duty cycle with a 15 minute recording duration in a 20 minute cycle (Table 1). Since there was continuous data from three other deployments at this site, the duty cycle was evaluated on continuously sampled data from the first deployment (less presence of sperm whales) and the fourth deployment (more presence of sperm whales). Random samples of the 15/20 duty cycle were taken from the deployments, shifting the listening period by one minute to find the proportion of overall recording effort. Sperm whale clicks in the 2017 PI deployment were linearly adjusted by 0.0839, resulting from the mean of forced duty cycles on the 2016 and 2019 deployments (0.0939 and 0.0738 respectively).

Changes in Sea Ice Concentration

A monthly gridded sea ice concentration (SIC) product ranging from 1850 to 2017 was used to evaluate historical changes of sea ice within a 20 km radius around the PI recording site (Walsh et al. 2019). Only data from 1901 and beyond was included to avoid data sets that had non-observed data which were supplemented with estimates. The 20 km radius was selected as a maximum detection range for sperm whales clicks based on a previous study (Tran et al. 2014) and distance within which most interactions with the surface would likely occur. Six ¼ degree latitude by ¼ degree longitude grid cells were within 20 km of the recording site and included in the analysis. This sea ice product merges 18 different data sources and provides a single mid-month day (MMD) average from each available source. The MMD average for the six grid cells near the recording site were averaged across the multiple sources (when available) to produce a single MMD average for the 20 km radius around the site. A linear regression was used to model the relationship between year and the median MMD SIC average by fitting a linear equation to the observed data in R Statistical Software (R Core Team 2013).

A finer resolution of mean daily SIC was used to compare with daily presence of sperm whales during our recording period from 2015-2019. Daily Advanced Microwave Scanning Radiometer 2 (AMSR2) sea ice maps were obtained from the University of Bremen (Spreen et al. 2008) and processed using Windows Image Manager (WIM) and Windows Automation Module (WAM) software (Kahru 2001) to produce an annual time series of mean daily SIC within a 20 km radius mask around the recording site. WAM software was used to compute the daily arithmetic mean, variance, and median of the SIC as a percent of the total mask area. The data excludes locations within 1 km from land to reduce edge effects and influence of snow on land.

Usage Notes

Please view the README.txt file for more information.