Claire Lacey

Large volumes of passive acoustic data are collected by researchers, governments, and conservationists across the globe to monitor species for population assessments and conservation objectives. An analysis bottleneck often exists due to the lack of tools to process large datasets and identify detected signals to the species level. Deep learning has been shown to enable fast and effective detection of species‐specific vocalizations. This study developed a Risso's dolphin ( Grampus griseus ) binary echolocation click classifier using a Long Short‐Term Memory neural network. This classifier was trained and validated on 872 visually verified single species encounters collected during towed hydrophone surveys, primarily in the Northeast Atlantic. The classifier was subsequently tested on an independent set of 31 h of visually verified data collected from static and drifting recorders in Scottish and English waters, comprising four different delphinid species (including Risso's dolphins) as well as various sources of anthropogenic noise. The best classifier achieved an F1 score of 0.98 (precision 0.96, recall 0.98). While care must be taken when applying deep learning classifiers to new datasets, the performance of this model on independent data highlights its ability to generalize well within the study region, even when using data collected on different recording platforms.

Spinner dolphins ( Stenella longirostris subsp.) in Hawaiʻi face the highest documented human activity exposure rates of any dolphin species worldwide, raising concerns about long‐term impacts. Fundamental to any appropriate management action for a protected species is the availability of reliable information on their abundance and distribution. Currently, abundance estimates are only available for one of the six designated stocks of spinner dolphins in Hawaiʻi. To address this information gap, small‐boat line‐transect surveys for spinner dolphins were conducted around the island of Oʻahu between October 2020 and May 2022. A total of 73 survey days were achieved, resulting in 3098 km of transects surveyed around the Oʻahu coastline. We used a novel method utilizing opportunistically collected unoccupied aerial system data to assess dolphin availability at the surface. We also investigated the potential for responsive movement of animals, for which no evidence was found. Estimated abundance of spinner dolphins for Oʻahu was 594 animals (95% CI [360, 980], CV = 0.26). Our findings represent the first circum‐island abundance estimate of Oʻahu's spinner dolphins, allowing management to explore strategies aimed at minimizing human‐dolphin interactions in the region and assess long‐term impacts.

Several legal acts mandate that management agencies regularly assess biological populations. For species with distinct markings, these assessments can be conducted noninvasively via capture‐recapture and photographic identification (photo‐ID), which involves processing considerable quantities of photographic data. To ease this burden, agencies increasingly rely on automated identification (ID) algorithms. Identification algorithms present agencies with an opportunity—reducing the cost of population assessments—and a challenge—propagating misidentifications into abundance estimates at a large scale. We explored several strategies for generating capture histories with an ID algorithm, evaluating trade‐offs between labor costs and estimation error in a hypothetical population assessment. To that end, we conducted a simulation study informed by 39 photo‐ID datasets representing 24 cetacean species. We fed the results into a custom optimization tool to discern the optimal strategy for each dataset. Our strategies included choosing between truly and partially automated photo‐ID and, in the case of the latter, choosing the number of suggested matches to inspect. True automation was optimal for datasets for which the algorithm identified individuals well. As identification performance declined, the optimization recommended that users inspect more suggested matches from the ID algorithm, particularly for small datasets. False negatives (i.e., individual was resighted but erroneously marked as a first capture) strongly predicted estimation error. A 2% increase in the false negative rate translated to a 5% increase in the relative bias in abundance estimates. Our framework can be used to estimate expected error of the abundance estimate, project labor effort, and find the optimal strategy for a dataset and algorithm. We recommend estimating a strategy's false negative rate before implementing the strategy in a population assessment. Our framework provides organizations with insights into the conservation benefits and consequences of automation as conservation enters a new era of artificial intelligence for population assessments.

Killer whales (Orcinus orca) have group‐specific call repertoires that can be used to track groups and populations using passive acoustic monitoring. To provide a detailed description of the Icelandic killer whale repertoire and its variation, we analyzed acoustic data collected in five locations between 1985 and 2016. Calls were classified manually, and CART and random forest analyses were employed to validate the manual classification. A total of 91 call categories (including call types and subtypes) were defined. Most call categories were recorded in more than one location, with the highest proportion shared between herring grounds in Vestmannaeyjar (South) and Breiðafjörður (West). However, both locations included call categories that were not recorded elsewhere in Iceland. Recordings from past herring wintering grounds in eastern Iceland included few call categories that matched other locations. Sample sizes from Reykjanes (Southwest) and Skjálfandi (North) were small and did not include unique call categories. The relative occurrence of call categories in Vestmannaeyjar changed little over a 14‐year period (2002–2016), although shorter‐term changes between years were observed that appeared to correlate to changes in individuals identified. This comparison of acoustic repertoires provides valuable information on the social structure and movement patterns of herring‐eating killer whales around Iceland.

A series of visual–acoustic surveys were carried out in the Mediterranean Sea between 2003 and 2007 from RV ‘Song of the Whale’. Almost 21,000 km of trackline were surveyed between the longitudes of 14°W and 36°E with an emphasis on regions with low survey effort. Survey tracklines were designed to provide even coverage probability with random start points. Ten cetacean species were positively identified (sperm whale, fin whale, Cuvier's beaked whale, false killer whale, long-finned pilot whale, Risso's dolphin, common bottlenose dolphin, rough-toothed dolphin, striped dolphin and short-beaked common dolphin). Several of these species, plus sei whale and harbour porpoise, were also encountered in the Atlantic contiguous area (the entrance waters of the Mediterranean between the Iberian Peninsula and north-west Morocco). These surveys expand and clarify the known distributions of cetaceans within the Mediterranean basin. New species documented from Libyan waters include sperm whale, striped dolphin and rough-toothed dolphin. False killer whales and rough-toothed dolphins were documented for the first time off Cyprus. Live harbour porpoises were seen for the first time on Morocco's Atlantic seaboard. It is suggested that the status of rough-toothed dolphins in the Mediterranean be revised from visitor to regular species. Substantial new information on encounter rates is now available for the planning of a basin-wide systematic survey of cetaceans within the Mediterranean Sea and contiguous Atlantic waters.

Photo-grammetric techniques developed for measuring body length and small scale movement patterns of cetaceans were applied to surface swimming basking sharks off the west coast of Scotland. These methods removed the need for close approaches, reducing the likelihood of disturbing the focal animal. Average swimming speed was calculated from the total path length between shark locations measured at approximately one minute intervals. These average speeds varied from 0.49 to 0.73 ms⁻¹ for tracks of between 30 and 170 minutes' duration. Body length measurements ranged between 2.35 and 6.43 m. For ten sharks where body length and swimming speed were measured there was a significant correlation between body length (L) and swimming speed (V) expressed as V = 0.36L⁰³³.

Current concerns about the population levels of the basking shark (cetorhinus maximus) in the north-east atlantic have prompted a need to understand population distribution, habitat preference and centres of abundance. in this study, spatial distribution maps derived from satellite-tag geolocations, boat surveys and public sightings data were compared. the broad distribution patterns revealed by these different methods are similar, but there are considerable differences in density distributions. surface sightings data show high densities, or ‘hotspots’ in the hebridean sea, clyde sea, irish sea and close inshore around devon and cornwall. tag geolocations, in contrast, identified two areas where individuals spent considerable time outside the distributions indicated by surveys and public sightings: the celtic sea and western approaches of the english channel. the reason for this disparity and its implications for population estimates for the species are discussed.