Task 4d

Investigating inter-click interval patterns for deep divers

Investigating inter-click interval (ICI) patterns will feed directly into considerations of factors driving cue rates such as time of day, season, geographic location, population or behavioral state. Odontocete studies have shown a modal distribution of inter-click intervals (ICIs), which also appear to be species-specific. While there generally is a preferred signaling rate, this may vary depending on several factors. We propose to investigate a comprehensive dataset for Cuvier’s and Blainville’s beaked whales as well as sperm whales with which analysis of ICI variability across time and space will be possible. Long-term datasets collected over multiple years will provide temporal, whereas data from multiple sites within a smaller geographic region will document variability across sites. 

Team members

• Jenny Trickey
• Simone Bauman-Pickering
• Natalie Posdaljian

Species Considered

• Blainville’s beaked whale (Mesoplodon densirostris)
• Goose-beaked whale (Ziphius cavirostris)
• Sperm whale (Physeter macrocephalus)

Main Results

Below we describe in turn the main results of each of the papers under task 4d

Geographic differences in Blainville’s beaked whale (Mesoplodon densirostris) echolocation clicks

Baumann Pickering et al. 2023

Data available

This study collected passive acoustic data from 76 locations across the North Pacific, Western North Atlantic, and Gulf of Mexico, spanning 150 cumulative years from 2004 to 2021. The data were collected using High-frequency Acoustic Recording Packages (HARPs; Wiggins & Hildebrand 2007). Data from this study is hosted on Dryad (Baumann-Pickering et al., 2013).

Figure 1. Deployment locations of HARPs (circles) in the Northern Hemisphere. Circle color and size indicate the percentage of days with Blainville’s beaked whale (Md) acoustic detections at that site. Light blue background shading indicates known range of this species based on the International Union for the Conservation of Nature (IUCN) Red List map (Pitman & Brownell Jr., 2020). Site names are only labeled for sites with Md detections.

Approach taken

This study involved a multi-phase process to detect and classify beaked whale signals. First, an automated click detector (Roch et al., 2011; Soldevilla et al., 2008) was used to identify echolocation clicks followed by a decision about the presence or absence of beaked whale signals based on a heuristically optimized expert system (Baumann-Pickering et al., 2013). Then an unsupervised learning strategy following Frasier et al., 2017 was used to analyze Md FM pulse signals, focusing on eliminating false detections and documenting signal variability across sites. In Step 1, clustering was applied to group and filter out false detections using spectral shape and interclick interval (ICI) distributions. Signals were first grouped into clusters within 5-minute time bins and then further refined using the Chinese Whispers algorithm. In Phase 2, similar clusters were identified across the deployment period, with spectral and ICI similarities used to optimize the clusters. In Step 2, clustering was repeated with adjusted parameters to improve the differentiation of beaked whale signal types, followed by manual screening to eliminate any remaining false signals. Sites were categorized based on the frequency of Md echolocation click detections, and a latitudinal cline in signal peak frequency was noted for sites with regular occurrences. Summary statistics for each step were documented, including cluster counts and signal characteristics.

Results

The analysis revealed geographic differences in echolocation click peak frequencies, with a notable latitudinal gradient: higher peak frequencies were detected in lower latitudes, and lower frequencies were found in higher latitudes (Figure 2). This gradient suggests potential acoustic differentiation between populations of Blainville’s beaked whale.

Figure 2. Increasing median peak frequency (with 25th and 75th percentiles) with lower latitudes (red regression line fit to medians: y = -4.8x + 191.7, R2 = 0.93) at sites with regular Blainville’s beaked whale presence (>1% of recording days).

When comparing ICI across sites, the median ICI of dominant clusters ranged from 270 to 390 ms (Figure 3). However, across regions, there were no clear differences in median ICI values, meaning ICI did not show the same latitudinal cline noted for peak frequency. The greatest variability within a region occurred in the Atlantic (70 ms), whereas in the Pacific, median ICI values were within 35 ms for all regions (Figure 3).

Figure 3. Interclick interval (ICI) of echolocation clicks in primary cluster at all sites with regularly occurring Blainville’s beaked whale FM pulse type detections (>1% of recording days). Circle indicates median, bars show 25th and 75th percentile, lines represent most extreme data points, while outliers are excluded. Sites are grouped according to region. Grey line indicates 300 ms ICI for visual orientation.

Conclusions

This study supports that it may be possible to acoustically delineate populations of Blainville’s beaked whales. The documented negative correlation between signal peak frequency and latitude could relate to body size. Body size has been shown to influence signal frequency, with lower frequencies produced by larger animals, which are subsequently more common in higher latitudes for some species, although data are lacking to adequately investigate this for beaked whales. Prey size and depth may shape the frequency content of echolocation signals, and larger prey items may occur in higher latitudes, possibly resulting in lower signal frequencies of their predators.

Passive Acoustic Density Estimation of Goose-Beaked Whales in Southern California Waters

Baumann-Pickering et al. (in prep)

Data Available

This project utilized passive acoustic long-term data collected from 2007 to 2020 in Southern California waters to estimate the density of Goose-beaked whales (Figure 4). Data was gathered from 25 different sites, with recording efforts ranging from 70 days to several years depending on the location.

Figure 4. Passive acoustic monitoring sites used for goose-beaked whale density estimation in Southern California. 

Approach Taken

Two density estimation approaches using distance sampling-based methods (Marques et al., 2013) were applied: click counting (Marques et al. 2009) and bin counting (Hildebrand et al. 2015). The analysis followed several key steps:

1. Click Detection: Acoustic signals (clicks) from goosebeaked whales were automatically detected from hydrophones at various recording sites with analyst-assisted software (Baumann-Pickering et al., 2013).
2. Click and time bin counting: Clicks and time bins containing these clicks were counted. Each positive time bin assumed a group of whales was present, translating these detections into a number of animals.
3. Multipliers: Variables such as group size, cue rate (click rate), and detection probability of goose-beaked whales were applied to adjust the results accordingly.

Results

This study calculated the density of goose-beaked whales at multiple locations, with significant variation across sites (Figure 5). Reported results are preliminary and multipliers are in the process of being optimized. The highest density was seen at site E (3.7 +/- 0.4 animals/100 km^2 (Figure 5). The cue rate was calculated using the inverse of the inter-click interval (ICI) and the proportion of time that whales spent clicking during dives (Figure 6). For example, at Site E, the cue rate was calculated as 0.53 clicks per second based on the mean ICI for the site (460 ms) and the proportion of a dive the animals spend click (24%; derived from tag data) (Figure 6).

Figure 5. Density estimates across different monitoring sites with notable variability in goose-beaked whale presence.

Figure 6. Inter-click interval (ms) of goose-beaked whales at the different recording sites. Rows represent locations in the Southern California Bight (inshore, central bight, shelf break, offshore) and the density of animals (low, medium, high). The median value of inter-click interval is in blue in the top right-hand corner of the plots. 

Conclusions

This analysis provided valuable insights into the spatial and temporal presence of goose-beaked whales in Southern California waters over 13 years. The study highlighted considerable variability in whale density across different monitoring sites, with site E showing the highest density. One of the key components for refining density estimates was the calculation of site-specific cue rates, which relied on the inter-click interval (ICI)—the time between successive echolocation clicks—and the proportion of time whales spent clicking during dives. By using site-specific ICIs, the project translated detected clicks into site-specific estimates of animal density, addressing a core challenge of density estimation from passive acoustic monitoring. These results demonstrate the importance of long-term acoustic monitoring to understand the behavior and distribution of elusive species like goose-beaked whales. The study also underscored the need for ongoing monitoring to track trends over time and adapt to environmental changes.

Demographic Specific Density Trends for Sperm Whales in the Western North Atlantic

Posdaljian et al. (in prep)//Posdaljian 2023 (PhD Dissertation; Chapter 3)

Data available

The study utilized data collected from 12 sites in the Western North Atlantic (WNA) between 2015 and 2019, using High-frequency Acoustic Recording Packages (HARPs; Wiggins and Hildebrand 2007). This comprehensive dataset encompassed nearly 36 years of cumulative recordings across these sites, providing extensive information on sperm whale echolocation patterns. The geographic distribution of the data was divided into northern and southern regions relative to the Gulf Stream (Figure 7).

Figure 7. The western North Atlantic study region with HARP locations represented by circle markers and labeled with the site abbreviation. Bathymetry is represented with a blue color scale and is given in meters. The inset map shows the two HAT locations. Site abbreviations: Heezen Canyon – HZ, Oceanographer Canyon – OC, Nantucket Canyon – NC, Babylon Canyon – BC, Wilmington Canyon – WC, Norfolk Canyon – NFC, Hatteras – HAT, Gulf Stream – GS, Blake Plateau – BP, Blake Spur – BS, Jacksonville – JAX . The black line represents the division between the northern and southern regions as referred to in this study as “North” and “South”, respectively. The shaded regions represent the four Western North Atlantic regions: Georges Bank, Southern New England, Mid-Atlantic Bight, and South-Atlantic Bight.

Approach taken

The detection of sperm whale clicks involved a multi-step approach described in Solsona-Berga et al., (2022) to ensure that only relevant clicks were analyzed, excluding potential noise from ship propellers or other sources of interference. The demographic classification of sperm whales was based on inter-click interval (ICI), a reliable metric for distinguishing between different animal sizes and hence demographic classes of sperm whales (Solsona-Berga et al., 2022). Click characteristics were used to categorize the whales into three groups: Social Groups (ICI ≤ 0.6 s), Mid-Size animals (0.6 s < ICI < 0.8 s), and Adult Males (ICI ≥ 0.8 s) (Solsona-Berga et al., 2022, Posdaljian et al., 2024).

Two density estimation methods were used to quantify sperm whale populations: click counting (Marques et al., 2009) and group counting (Hildebrand et al., 2015). The click counting method estimated density based on individual echolocation clicks detected at each site, while the group counting method calculated densities from the presence of echolocation clicks within 5-minute time bins (Hildebrand et al., 2015), which indicated the presence of groups of whales. Both methods require integrating the probability of detecting acoustic cues, vocal activity (proportion of time spent clicking versus time spent silent), and group size that was either derived from in-situ data, existing literature, or modeled. 

Distance sampling-based methods using a Monte Carlo simulation approach have been used to estimate the probability of detecting marine mammal sounds from a fixed single sensor (Küsel et al. 2011, Helble et al. 2013, Frasier 2016, Hildebrand et al. 2019, Solsona-Berga et al. 2024). This requires an understanding of specific aspects of sperm whale behavior and their acoustic environment. These models incorporate factors like echolocation signal characteristics, sound propagation, animal behavior, and receiver sensitivity, drawing from previously collected data or relevant literature. Simulations were tailored to account for dive behavior, click characteristics, and body orientation of the whales, with parameters derived from observed data when possible or based on modeled data if direct observations were unavailable. 

Understanding sperm whale vocal activity is essential for estimating animal densities. The proportion of time whales spend clicking versus silent during foraging dives was calculated using data from tag studies. For Social Groups and Mid-Size whales, about 80.7% of their time in foraging dives is spent clicking, while Adult Males spend around 91% of their time clicking (Watwood et al., 2006, Teloni et al., 2008). Cue rates, which indicate how often whales produce clicks, were simulated by considering dive parameters and inter-click intervals (ICI). The simulation accounted for variations in whale behavior and produced a distribution of cue rates used for density estimation. For group size estimation, data from visual and aerial surveys conducted by NOAA were used. These surveys provided estimates of group sizes, with adjustments made based on sightings and follow-ups. A minimum group size threshold of three was set to distinguish between solitary males or juvenile groups and larger Social Groups. For Adult Males, a group size of 1.5 was assumed to account for occasional pairs, while a group size of 2 was assumed for Mid-Size groups, likely comprising juvenile males or larger females.

Results

All detailed results from this study can be found in Posdaljian (2023). Particularly interesting for the ACCURATE project were differences in ICI and as a result, cue rate. Using the dive parameters from Watwood et al. (2006) and Teloni et al. (2008), the model-estimated proportion of time spent clicking was 58% for Social Groups/Mid-Size and 66% for Adult Males. These estimates were divided by the class specific ICI to estimate site-specific cue rates for click counting. ICI distributions were relatively consistent across sites, with Social Groups and Mid-Size whales showing a mean ICI of 494 ms and 642 ms, respectively (Figure 8). Adult Males had a more variable ICI at 864 ms (Figure 8). These ICI values were used to estimate cue rates, which indicate the frequency of clicks. Social Groups had the highest cue rate at 1.29 clicks per second, followed by Mid-Size whales at 1.00 clicks per second, and Adult Males at 0.81 clicks per second (Table 1).

Density estimates for sperm whales were calculated using both click counting and group counting methods. The two approaches showed good agreement, with Social Groups being the dominant size class, accounting for over 80% of the population in the region (Figure 9, Table 2-3). Mid-Size whales were the second most common, while Adult Males were rare, making up less than 1.5% of the population (Table 2-3).

Spatial analysis revealed that all size classes had higher densities at northern sites, with the highest density observed at site NC. The southern sites showed significantly lower densities, indicating a clear spatial variation in sperm whale distribution (Table 2-3).

Long-term density trends were mainly observed in Social Groups, with increasing densities noted at northern sites (Table 4). The most significant annual increases were observed at site NFC, with smaller increases at other northern sites (Table 4). In contrast, the southern sites showed either stable or declining densities for Social Groups. Other size classes did not show a discernible trend in density due to their lower numbers.

Figure 8. Figure S13. Violin plots reveal interclick interval distribution for each class at each site. The white marker represents the mean of the distribution, and the black line represents the peak of the GMM distribution fit to the data.

Table 1. Table S9. Cue rate (clicks/s) and CV for each class (column) at each site (row).

Figure 9. Weekly density estimates for Social Groups at the northern sites based on the click (left) and group (right) counting approaches. The circles denote mean density estimates, the vertical lines represent plus and minus one standard error, and shaded areas show gaps in recording effort. Note that each subplot features a different y-axis scale to accommodate the varying ranges of the data.

Table 2. Average sperm whale densities derived from click counting for each class (column) at each site (row) given in # of animals per 1000 km2 ± standard deviation. 

Table 3. Average sperm whale densities derived from group counting for each class (row) at each site (column) given in # of animals per 1000 km2 ± standard deviation.

Table 4. Annual trends in click % (left) and group % (right) change per year for Social Groups at all sites (rows) with the 95% confidence intervals represented by the respective minimum and maximum values.

Conclusion

This study sheds light on the demographic-specific acoustic density of sperm whales in the Western North Atlantic (WNA) by analyzing data from 12 sites over four years. It focused on three groups: Social Groups (females and their young), Mid-Size animals, and Adult Males. Results showed that sperm whale presence was significantly higher in the northern sites, especially north of the Gulf Stream, with Social Groups being the most prevalent. The increase in Social Group density in the northern region suggests favorable ecological conditions. The study highlights the importance of considering demographic differences in sperm whale density estimates and the need for targeted conservation strategies that address specific sex and age groups to ensure the effective management of sperm whale populations in this crucial oceanic ecosystem.

References

Baumann-Pickering, S., McDonald, M.A., Simonis, A.E., Solsona Berga, A., Merkens, K.P., Oleson, E.M., Roch, M.A., Wiggins, S.M., Rankin, S., Yack, T.M. and Hildebrand, J.A., 2013. Species-specific beaked whale echolocation signals. The Journal of the Acoustical Society of America, 134(3), pp.2293-2301.

Baumann‐Pickering, S., Trickey, J.S., Solsona‐Berga, A., Rice, A., Oleson, E.M., Hildebrand, J.A. and Frasier, K.E., 2023. Geographic differences in Blainville’s beaked whale (Mesoplodon densirostris) echolocation clicks. Diversity and Distributions, 29(4), pp.478-491.

Baumann-Pickering, Simone; Trickey, Jennifer S.; Solsona Berga, Alba et al. (2022). Blainville’s beaked whale (Mesoplodon densirostris) echolocation clicks from autonomous passive acoustic recordings [Dataset]. Dryad. https://doi.org/10.6076/D12G6N

Frasier, K.E., Roch, M.A., Soldevilla, M.S., Wiggins, S.M., Garrison, L.P. and Hildebrand, J.A., 2017. Automated classification of dolphin echolocation click types from the Gulf of Mexico. PLoS computational biology, 13(12), p.e1005823.

Frasier, K.E., Wiggins, S.M., Harris, D., Marques, T.A., Thomas, L. and Hildebrand, J.A., 2016. Delphinid echolocation click detection probability on near-seafloor sensors. The Journal of the Acoustical Society of America, 140(3), pp.1918-1930.

Helble, T.A., D’Spain, G.L., Hildebrand, J.A., Campbell, G.S., Campbell, R.L. and Heaney, K.D., 2013. Site specific probability of passive acoustic detection of humpback whale calls from single fixed hydrophones. The Journal of the Acoustical Society of America, 134(3), pp.2556-2570.

Hildebrand, J.A., Baumann-Pickering, S., Frasier, K.E., Trickey, J.S., Merkens, K.P., Wiggins, S.M., McDonald, M.A., Garrison, L.P., Harris, D., Marques, T.A. and Thomas, L., 2015. Passive acoustic monitoring of beaked whale densities in the Gulf of Mexico. Scientific reports, 5(1), p.16343.

Küsel, E.T., Mellinger, D.K., Thomas, L., Marques, T.A., Moretti, D. and Ward, J., 2011. Cetacean population density estimation from single fixed sensors using passive acoustics. The Journal of the Acoustical Society of America, 129(6), pp.3610-3622.

Marques, T.A., Thomas, L., Ward, J., DiMarzio, N. and Tyack, P.L., 2009. Estimating cetacean population density using fixed passive acoustic sensors: An example with Blainville’s beaked whales. The Journal of the Acoustical Society of America, 125(4), pp.1982-1994.

Posdaljian, N., Solsona-Berga, A., Hildebrand, J.A., Soderstjerna, C., Wiggins, S.M., Lenssen, K. and Baumann-Pickering, S., 2024. Sperm whale demographics in the Gulf of Alaska and Bering Sea/Aleutian Islands: An overlooked female habitat. Plos one, 19(7), p.e0285068.

Roch, M.A., Klinck, H., Baumann-Pickering, S., Mellinger, D.K., Qui, S., Soldevilla, M.S. and Hildebrand, J.A., 2011. Classification of echolocation clicks from odontocetes in the Southern California Bight. The Journal of the Acoustical Society of America, 129(1), pp.467-475.

Soldevilla, M.S., Henderson, E.E., Campbell, G.S., Wiggins, S.M., Hildebrand, J.A. and Roch, M.A., 2008. Classification of Risso’s and Pacific white-sided dolphins using spectral properties of echolocation clicks. The Journal of the Acoustical Society of America, 124(1), pp.609-624.

Solsona-Berga A, Frasier KE, Posdaljian N, Baumann-Pickering S, Wiggins, S., Soldevilla, M., Garrison, L., Hildebrand, J.A. 2024. Accounting for sperm whale population demographics in density estimation using passive acoustic monitoring. Mar Ecol Prog Ser 746:121-140. 

Solsona‐Berga, A., Posdaljian, N., Hildebrand, J.A. and Baumann‐Pickering, S., 2022. Echolocation repetition rate as a proxy to monitor population structure and dynamics of sperm whales. Remote Sensing in Ecology and Conservation, 8(6), pp.827-840.

Teloni, V., Mark, J.P., Patrick, M.J. and Peter, M.T., 2008. Shallow food for deep divers: Dynamic foraging behavior of male sperm whales in a high latitude habitat. Journal of Experimental Marine Biology and Ecology, 354(1), pp.119-131.

Watwood, S.L., Miller, P.J., Johnson, M., Madsen, P.T. and Tyack, P.L., 2006. Deep‐diving foraging behaviour of sperm whales (Physeter macrocephalus). Journal of Animal Ecology, 75(3), pp.814-825.
Wiggins, S.M. and Hildebrand, J.A., 2007, April. High-frequency Acoustic Recording Package (HARP) for broad-band, long-term marine mammal monitoring. In 2007 symposium on underwater technology and workshop on scientific use of submarine cables and related technologies (pp. 551-557). IEEE.

Progress

This task is now complete. The manuscripts in preparation will at some point become published papers, but that will probably only happens ater the project’s life time.

Watch this space for task updates.