Written by Lars Bejder
Members of the MMRP are thrilled to be attending the 25th Biennial Conference on the Biology of Marine Mammals in Perth, Western Australia in November 2024. We are delighted to have the opportunity to hear about all the incredible research that is occurring all over the world, to learn from peers, and to contribute to the many pressing conservation issue facing of marine mammals globally.
We are also looking forward to sharing our research with colleagues, friends, peers, NGOs, managers, industry, and community stakeholders. We hope our work will stimulate discussions and that we will receive feedback such that we can optimize the conservation outcomes of our efforts. Specifically, we will be presenting four poster, four speed talks, five 12-min oral presentations, and a full-day workshop. Below, is an overview of all of our first-authored presentations that we will be reporting on – including abstracts. Please reach out if you have any questions or would like to hear more. See ya’ down-under!!
Posters
Title: Talking story with ʻīlio holo i ka uaua (Hawaiian monk seals): First descriptions of underwater sound production in free-ranging individuals
Authors: Kirby Parnell, Kyleigh Fertitta, Adriana Diaz, Caroline Smith, Pearl Thompson, Phil Patton, Isabelle Charrier, Michelle Barbieri, Aude Pacini and Lars Bejder.
Abstract: The Hawaiian monk seal (HMS; Neomonachus schauinslandi), known as 'īlio holo i ka uaua, is an endemic and endangered species with an estimated population of 1,600 individuals. Research efforts over the last 40 years have resulted in a wealth of knowledge in HMS biology, movements, and population ecology, but its acoustic behavior remains largely unknown. HMS underwater vocal behavior has only been described for two individuals living in human-care. To broaden our understanding of sound production for free-ranging wild seals, we deployed passive acoustic recorders at four shallow-water critical habitats throughout the Hawaiian Archipelago. In 2,576 hours of acoustic recordings, we manually detected and classified >21,000 underwater vocalizations. A discriminant function analysis based on 13 acoustic parameters (e.g., duration, peak frequency) was performed on seven call types and resulted in an average correct classification rate of 87%. We found four of six previously-described call types from seals in human-care (croak, growl, moan, whoops) and three new call types (foraging call, moan growl, and moan foraging call). All vocalizations are low-frequency (<1 kHz), short-medium duration (<7 sec), and always produced in bouts of more than one call. The newly described foraging call was also documented in 11 citizen scientists’ videos showing seals simultaneously foraging and calling. Call detection rates were significantly lower at Mānana (average 0.79 calls per hour; SE=0.83; P<0.001), versus Lehua Rock, Manawai, and Lalo (average 26 calls per hour; SE=0.83). There were no clear diel patterns in sound production at any site. This first description of the underwater vocal behavior in free-ranging wild HMS highlights the significance of vocal activity for this endangered species.
Title: Low Energetic Expenditure of Short-Finned Pilot Whales Uncovered From Dive Behavior, Swimming Kinematics, and Morphometrics
Authors: William Gough, Brijonnay Madrigal, Jens Currie, Lewis Evans, Martin van Aswegen, Stephanie Stack, Aude Pacini and Lars Bejder.
Abstract: Diving is one of the most energetically demanding behaviors undertaken by marine mammals, with many species relying on large body size to reduce energetic expenditure in relation to prey intake. Short-finned pilot whales (Globicephala macrorhynchus) are one of the smallest cetacean species that forage at great depths (~1000m). We deployed biologgers (Customized Animal Tracking Solutions) equipped with cameras, hydrophones, and tri-axial accelerometers on short-finned pilot whales (n=6; 56 data hours) in Hawaiʻi. From this dataset, we measured dive-scale metrics: maximal depths (510±207 m), fluking cessation depths (40.7±9.6 m on descent; 31.5±5.8 m on ascent), and duration (10.6±4.0 mins) of foraging dives (n=111). Animals displayed similar v-shaped foraging dives to those found by previous studies. We found that animals spent ~4% of their time in the bottom phase of these dives, with 1-4 prey-capture attempts per dive (estimated from accelerometry and acoustic data). Subsequently, we integrated body morphometrics (including body length and tail planar area) derived from UAS-photogrammetry data collected concurrently with orientation data from our biologgers. This enabled us to quantify key kinematic and hydrodynamic parameters of swimming: average oscillatory frequency (0.56±0.05 Hz), forward speed (1.74±0.39 ms-1), mass-specific thrust power (0.29±0.09 W kg-1), propulsive efficiency (84±2%), and percentage of time spent actively fluking (40±12%). From these values, we calculated the energetic cost of swimming per second (0.04 ± 0.02 kJ s-1) and per day (1488±784 kJ day-1). Combining these costs with basal metabolic rates (25,733±8,395 kJ day-1; derived as 70*mass-75 from allometric equations of body mass), we estimated the number of prey items - such as squid (~6-11 at 150g each) - necessary to cover daily energetic expenditure. Our results suggest that, despite their smaller size, energy expenditure remains low for short-finned pilot whales, allowing them to subsist on few prey items.
Title: Descriptions of Hawaiian monk seal underwater vocal behavior at Lehua Rock
Authors: Kyleigh Fertitta, Adriana Diaz, Caroline Smith, Phil Patton, Kirby Parnell, Isabelle Charrier, Aude Pacini and Lars Bejder.
Abstract: The underwater vocal behavior of the endangered Hawaiian monk seal (HMS; Neomonachus schauinslandi) was previously described for two seals in human-care who produced six call types. However, there is a dearth of information pertaining to the vocal behavior of free-ranging HMS. We recorded vocalizations of wild HMS for seven days in May 2021 at Lehua Rock, Niʻihau which is a popular destination for boat-based ecotourism (e.g., snorkeling, scuba diving) during the spring and summer months. Using a continuously-recording SoundTrap 500HF, we analyzed passive acoustic data to 1) detect and categorize HMS vocalizations and 2) describe diel patterns in sound production. Spectrograms were audio-visually inspected for five minutes per hour for a total of 12 h. Seals produced four of the six previously-described call types from seals in human-care: croaks (n=1921), growls (n=2877), and whoops (n=644) were most commonly recorded, moans (n=121) were rarely detected, and rumbles and groans were not detected. Additionally, three new call types were identified: foraging call, moan growl, and moan foraging call. Call detection rates were lowest between 1000-1300 and at 2000 which coincided with increased anthropogenic (ecotourism boats and scuba divers) and biological (vertically migrating layer) activity, respectively. These results provide evidence that Lehua Rock is an important habitat for HMS acoustic communication. Future studies should aim to investigate the impacts of anthropogenic and biological sounds on HMS vocal behavior at this designated critical habitat.
Title: Insights into 'īlio holo i ka uaua (Hawaiian monk seal): Exploring underwater vocalizations and behavioral contexts via Crittercams
Authors: Pearl Thompson, Kirby Parnell, Isabelle Charrier, Kenady Wilson, Aude Pacini and Lars Bejder.
Abstract: Animal-borne biologging instruments provide information on the fine-scale movement and behavior of free-ranging wild animals. National Geographic “Crittercams” have been used to elucidate the behavior of multiple species, including the Hawaiian monk seal (HMS; Neomonachus schauinslandi). Previously, these data were used to explore foraging behavior and prey interactions, but the underwater vocal behavior of this species remains largely unstudied. To describe the behavioral context of HMS underwater vocalizations, we analyzed simultaneous acoustic and video data recorded by Crittercams from seven male HMS instrumented in the Main Hawaiian Islands from 2010-2012. Within 9.5 hours of acoustic data, 1,471 underwater vocalizations and seven discrete call types were detected and classified, one of which is previously undescribed. When calls were detected by visualizing them on spectrograms, corresponding video data were reviewed to document seal behavior(s) and context of calls (i.e., solitary or multiple seals). A total of 2,745 behaviors were documented and classified using an ethogram of 12 behaviors. The most frequently observed behaviors while vocalizing were “Search” at 39%, “Swimming” at 34%, and “Lounging” at 14%. When alone, seals primarily produced Growls (35%), whereas in the presence of another seal, they most frequently produced Whoops (60%). The previously undescribed call type was detected during all documented “Foraging” behaviors and not detected otherwise. While limited to seven individuals, this study is the first attempt to document HMS behavior in relation to underwater sound production. Results indicate that most call types are produced in numerous behavioral contexts, suggesting the importance of HMS vocalizations in various activities.
Speed Talks
Title: Maternal investment, body condition and calf growth in humpback whales across Hawaiʻi and Southeast Alaska
Authors: Martin van Aswegen, Andrew Szabo, Jens Currie, Stephanie Stack, Lewis Evans, Jan Straley, Janet Neilson, Christine Gabriele, Kelly Cates, C. Scott Baker, Debbie Steel and Lars Bejder.
Abstract: Efficient energy allocation is crucial for maximizing fitness of migrating animals, with the cost of reproduction a key, yet understudied parameter in bioenergetic models. Despite recent declines in North Pacific humpback whale (Megaptera novaeangliae) reproductive output and calf survival, there is limited understanding of how mother-calf pairs allocate energy resources across the migratory cycle. Here, unoccupied aerial system (UAS; drone) photogrammetry was used to quantify the body size and condition of humpback whales on their Hawaiʻi (HI) breeding and Southeast Alaska (SEAK) feeding grounds, totaling 2431 measurements of 1669 individuals between 2019 and 2022. Repeated measurements of body volume (BV) and length (BL) were used in linear mixed models (LMMs) to quantify mean rates of change in maternal BV and calf growth. In HI, maternal BV decreased by -0.106 m3 day-1 (SD=0.013, n=118), while calf BV and BL increased by 0.035 m3 day-1 (SD=0.006, n=119) and 2.6 cm day-1 (SD=0.30), respectively. In SEAK, maternal BV increased by 0.015 m3 day-1 (SD=0.006, n=21) while calf BV and BL increased by 0.039 m3 day-1 (SD=0.008, n=21 calves) and 0.93 cm day-1 (SD=0.61), respectively. Maternal investment in calf growth correlated with both female body length and condition, with larger females producing larger, faster-growing calves. Finally, we used a LMM to compare intra-seasonal changes in body condition of pregnant (n=26), lactating (n=87), and unclassified females (n=93) in SEAK. Over four 150-day feeding seasons, lactating female body condition increased by an average of 5.10%, three times lower than unclassified females (15.51%) and six times lower than pregnant females (31.32%). These findings represent novel insights into the life history of humpback whales across their migratory cycle, providing key baseline data for bioenergetic models. A better understanding of the cost of reproduction is critical in elucidating the effects of extrinsic factors (e.g., anthropogenic disturbance and rapidly changing ocean ecosystems) on reproductive success.
Title: The energy dilemma: how the energetic demands of false killer whales and reduced prey availability may be contributing to fishery depredation
Authors: Jens Currie, Brian Stirling, Grace Olson, Stephanie Stack, Shannon Barber-Meyer, Nozomi Kobayashi, Suguru Higa, Keiichi Ueda, Andreas Fahlman, Herman Pontzer, Austin Allen, William Gough, Martin van Aswegen, Lewis Evans and Lars Bejder
Abstract: The failure of the endangered Main Hawaiian Islands insular distinct population segment of false killer whales (FKWs, Pseudorca crassidens) to rebound in numbers following its Endangered Species Act listing underscores a pressing conservation concern. The population size has remained low, with estimates hovering at only 130-150 individuals since 2012. Concurrent threats, including bycatch in fishing gear and the depletion of large prey species, collectively impact the population's health and potential for recovery. Using aerial photogrammetry, we detected changes in body condition over a five-year period that revealed instances of nutritional stress and prompted an investigation into the energetic costs inherent to FKWs in Hawaiʻi. Ongoing data collection from FKWs in human care provided preliminary data to quantify metabolic rates to determine energetic expenditure for rest and locomotion via flow meter respirometry and doubly labelled water measurements. These findings were then extended to their free-ranging counterparts in Hawaiʻi, who were tagged with non-invasive tri-accelerometer inertia tags to determine daily activity budgets. Combining these data, we were able to calculate the energetic expenditure associated with swimming strokes in wild FKWs and estimate their daily total energy expenditure. Preliminary results suggest that sustained high swim speeds (3-5 m/s), movement patterns spanning over 100 km/day, and observations of high foraging rates indicate a significant caloric need. This underscores the critical importance of maintaining sufficient prey levels to support the population's energetic demands. By understanding the energetic needs of FKWs in conjunction with trends in commercial fisheries catch data of preferred prey, we can better understand whether the observed high depredation rates are driven by their substantial energetic demand. Such information will help advise on the ongoing challenges of high fisheries interactions and will be instrumental in formulating precise management strategies for the conservation of this endangered population.
Title: Quantifying the abundance and survival rates of island-associated spinner dolphins using a multi-state open robust design model
Authors: Liah McPherson, Janelle Badger, Kyleigh Fertitta, Madison Gordanier, Cameron Nemeth and Lars Bejder.
Abstract: Spinner dolphins (Stenella longirostris subsp.) occupy the nearshore waters of several Hawaiian Islands. Due to their constrained behavioral pattern and genetic isolation, they are vulnerable to anthropogenic threats. Their occurrence and behavior are well-described, yet a lack of data on their abundance and survival rates hinders optimal conservation action. Using designed-based photo-identification surveys, this study estimated the abundance, apparent survival, and emigration of spinner dolphins off the Waiʻanae Coast of Oʻahu through multi-state open robust design (MSORD) and POPAN modelling. Eight seasonal field seasons, (two winter, spring, summer and autumn) each comprised of six surveys of the study area, were completed during two consecutive years. Seasonal abundance estimates derived from the best fitting model ranged from 140 ( 36.8 SE, 95% CI: 84-232) to 373 ( 60.0, 95% CI 273-509) individuals and were lowest during winter seasons. The MSORD estimated a survival rate of 0.95 (0.02 SE) and a Markovian pattern of temporary emigration. POPAN modelling estimated a super-population size of 627 ( 78 SE, 95% CI: 492-798), reflecting the total number of individual dolphins that used the study area during the entire study period. Additional research on circum- and inter-island dolphin movements around and between Oʻahu and the Maui Nui region may shed light on both seasonal movement patterns and overall abundance for the Oʻahu/4-Islands stock. This work represents the first systematic mark-recapture effort to assess the abundance and survival rates of these highly exposed dolphins, providing valuable insights for conservation and management.
Title: Increasing re-sighting rates of humpback whales: unoccupied aerial systems increase identification rates by > 225% when compared to traditional fluke identification
Authors: Lewis Evans, Martin van Aswegen, Sonja Feinberg, Jens Currie, Stephanie Stack and Lars Bejder.
Abstract: Photo-identification (photo-ID) provides crucial information on demographic parameters, movement patterns, and social structure of marine mammals. Humpback whales (HBWs) (Megaptera novaeangliae) are individually identified through distinguishing marks on the ventral side of their flukes. However, on the Hawai’i breeding ground, only 39% of HBW lactating females “fluke-out” (van Aswegen et al. in prep.), preventing identification and the ability to obtain within-season resights. Consequently, opportunities to document known individuals repeatedly within a season are lost, significantly reducing our ability to study their life history and biology. This study implemented a novel technique utilizing aerial videography and partially automated recognition software (I3S Classic) to establish a dorso-perspective identification catalog of HBWs. The catalog relies on individually identifiable features obtained from a top-down perspective via unoccupied aerial systems (UAS; drones): cookiecutter shark (Isistius spp.) scarring (CCS) and tubercle nodule patterns (TB). During encounters with 1520 HBWs (including repeats) between Jan-Mar 2022, we obtained photo-ID images of 785 flukes of sufficient quality to identify individuals. Using UAS-imagery and the I3S Classic algorithm, we assigned 1415 encounters with an aerial identification image (an increase of 80.25% compared with fluke IDs only). Our results show that CCS and TB are effective aerial identification markings, visible for at least 55 days, the longest resight period in the present study. Our top-down photo-ID catalog provided a 227% and 33.6% increase in resightings of known lactating females and mature adults, respectively. By expanding our methods of tracking individuals beyond fluke identification, we are better able to monitor within season trends (e.g. changes in body condition and residency times) and, when fluke identification is obtained, link this to critical life history information. Improved monitoring of individuals over spatial and temporal scales will better our understanding of habitat use, migratory timing and physiological change during vulnerable fasting and lactation periods.
Full Talks
Title: Quantifying acoustic behavior of false killer whales and short-finned pilot whales in Hawaiʻi via non-invasive bio-logging devices
Authors: Brijonnay Madrigal, Lars Bejder, William Gough, Jens Currie and Aude Pacini.
Abstract: False killer whales (FKW; Pseudorca crassidens) and short-finned pilot whales (SFPW; Globicephala macrorhynchus) are highly social, Hawaiʻi resident species which are vulnerable to anthropogenic activities (e.g. naval sonar, fishing interactions), therefore learning more about their acoustic behavior is imperative to conservation. While previous studies have used satellite telemetry tags to track movement, distribution, and diving behavior, they provide little information on behavioral context related to acoustic communication and foraging behavior. This study marks the first-time concurrent audio/video tags have been deployed on odontocetes in Hawaiʻi and provides high-resolution insight into call function and foraging behavior. Suction cup multi-sensor Custom Animal Tracking Solutions (CATS) tags, containing a camera and hydrophones (SR 96 kHz), were deployed off Maui and Lānaʻi islands on endangered FKW (n = 2) and SFPW (n = 4). Cumulatively, 60 hours of acoustic data were recorded. We manually detected/classified calls using Raven Pro and features were extracted using PAMGuard ROCCA. Clicks were detected using an automated MATLAB click detector. Preliminary analysis revealed that FKW produced higher call rates than SFPW (33 and 9.5 calls/min, respectively). Tagged FKW produced rarely documented pulsed calls and repertoires consisted of at least seven previously undescribed stereotyped calls. Most calls were biphonic (simultaneous call + clicks), a phenomenon only known to occur in five odontocetes, and likely function in communication. These species occupy different ecological niches – depth data revealed SFPW foraging depths of 600-800m and FWK dives to 30m with occasional deep dives to 100m. The SFPW data revealed the presence of regular clicks and terminal buzzes indicative of prey capture events during foraging dives which were visually validated using a depth-triggered light. This study will allow us to describe behavioral states, communication function, and calculate cue rates to supplement passive acoustic monitoring efforts and inform management and conservation in Hawaiʻi.
Title: Innovative approach for inferring age-structure of dolphin populations: A case study of spinner dolphins off Hawaiʻi Island
Authors: Fabien Vivier, Nicholas Hofmann, Kristi West, Liah McPherson, Cameron Nemeth, Martin van Aswegen, Aude Pacini, Cormac Booth and Lars Bejder.
Abstract: Assessing population trends and demographics is vital for managing long-lived, slow-reproducing species. However, obtaining demographic data, e.g., age-structure, is challenging. Traditional monitoring methods often fail to quickly identify demographic shifts, emphasizing the need for innovative approaches. We used Unoccupied Aerial System (UAS)-photogrammetry to assess the age-structure of the spinner dolphin (Stenella longirostris) stock off the Kona coast of Hawaiʻi Island through opportunistic boat-based surveys in September 2021 and 2022. This stock, estimated at 631±60.1SE individuals (Tyne et al. 2014), has been a NOAA management priority since 2005 due to concerns about its long-term viability amid rising human activities. Encounter rates in 2021 averaged 211±22 SE individuals daily across 49 groups. We established the relationship between total body length (TL) and blowhole-to-dorsal-fin-distance (BHDF) using age estimates and body morphometric data collected from 34 stranded specimens, which allowed for the development of a length-at-age growth curve. We documented a strong linear relationship between TL and BHDF (R²=0.92). A Gompertz growth model, with an asymptotic length of 199.7 cm, proved optimal. We estimated TL for surfacing dolphins using UAS-measurements of BHDF through post-hoc video processing. Using stranding data, we categorized measured individuals into three age-classes: calves (0–2 years old), juveniles (>2–7 years old), and adults (>7+ years old), enabling the assessment of both group and stock age-structure. We assessed the stock’s age-structure using the three surveys with the highest counts (295, 271, and 267 individuals), totaling 179, 219, and 162 measurements, respectively. Age-structure estimates remained consistent across days (p>0.1), averaging 10.2% calves, 51.4% juveniles, and 38.4% adults. This study presents a novel and effective approach for assessing the age-structure of delphinid populations, however, understanding birth and mortality rates is essential before concluding on the stock’s status. This technique shows potential for enhancing the speed, affordability, and efficiency of monitoring delphinid populations.
Title: Effect of Pectoral Fin Morphology on Humpback Whale Turning Performance
Authors: Cameron Nemeth, William Gough, Frank Fish, Andrew Szabo, Hunter Warick and Lars Bejder.
Abstract: The marine environment is experiencing climate change-induced declines in prey species availability and predictability. Morphological adaptations that enhance foraging efficiency of predators are likely to become increasingly important. The humpback whale (Megaptera novaeangliae) exemplifies adaptability with its unique morphology, featuring enlarged control surfaces (i.e. pectoral fins, tail flukes) alongside distinctive feeding strategies. To test whether the humpback whale’s morphology directly impacts its ability to perform its foraging behaviors, we analyzed a unique behavior requiring enhanced turning capabilities: solitary bubble-net feeding. This prey-capture strategy involves forming a tightening net of bubbles, followed by a lunge up through the middle of the net to engulf krill. To quantify turning performance during net formation (n=15 nets), we combined positional, orientation and speed data from tri-accelerometer, animal-attached biologgers (Customized Animal Tracking Solutions) with photogrammetric measurements of the animals and the formed bubble-nets using calibrated unoccupied aerial systems equipped with custom-fitted altimeters (DJI Inspire-2). These whales exhibited centripetal accelerations (0.46 m/s2) at the highest end of the range previously found for the species (0.07-0.49 m/s2) as well as smaller turning radii (~4.6 m) than either the smaller minke (~5.6 m) or larger blue (~12 m) whale. Swimming speed was maintained or increased as turning radius decreased. This finding contradicts movements observed for other whales, as they commonly disengage propulsion to achieve tighter radius turns, limiting speed and acceleration out of the turn. We calculated a higher lift force for humpback whale pectoral fins (1534 N) compared to fin whales (195 N) and sperm whales (276 N), two species with smaller pectoral fins. Our findings are the first to use foraging data from free-swimming humpback whales to quantify the relationship between their unique morphology and turning capability, as previously documented in hydrodynamic models.
Title: Quantifying energy expenditure in humpback whale mother-calf pairs: insights from biologging tags and UAS-photogrammetry
Authors: Augusta Hollers, William Gough, Martin van Aswegen, Andrew Szabo, Lewis Evans, Jens Currie, Stephanie Stack, Ashley Blawas and Lars Bejder.
Abstract: Quantifying daily energy expenditure of cetaceans is an important step in modeling their population level responses to climate change and anthropogenic disturbance, especially for capital breeders when mothers are simultaneously fasting and investing significant amounts of energy into calf growth and development. Technological advances in biologging tags and unoccupied aerial system (UAS; drone) photogrammetry can overcome the limits of traditional captive respirometry methods, allowing for new insight into large cetacean energetics. In this study, we used 55 biologging suction-cup tag deployments on humpback whale mothers and calves (Megaptera novaeangliae), and body mass models from UAS photogrammetric measurements of 1,405 mother-calf pairs, to determine their daily energy usage while on their winter breeding ground in Hawai’i. Energy expenditure for each tagged animal was quantified in two ways: using the respiration rate and body mass to estimate field metabolic rates (FMR), and using a hydrodynamic cost of transport (CoT) model combining energy expended while fluking and while at rest. Estimates of calf energy expenditure were lower using the CoT method (152 MJ/day, 95% CI 127-178) than the FMR method (301 MJ/day, 95% CI 260-342). Estimates of mother energy expenditure were similar for both methods (FMR: 1255 MJ/day, 95% CI 625-1886, CoT: 1178 MJ/day, 95% CI 1052-1304). We converted the UAS-derived average body volume loss of lactating females into a daily energy loss estimate of 2632 MJ/day (SD = 318) using published estimates of blubber lipid and protein concentrations. We subtracted the 1255 MJ/day used in metabolism to estimate 1345 MJ/day of energy transferred through milk to the calf. Our findings provide a best estimate of humpback whale energy expenditure, and are crucial to future studies on energy transfer, metabolism, and the relationship between behavior and energy usage in large cetaceans.
Title: Evaluating tradeoffs between automation and bias in population assessments relying on photo-identification
Authors: Phil Patton, Krishna Pacifici, Robin Baird, Erin Oleson, Jason Allen, Erin Ashe, Aline Athayde, Charla Jean Basran, Elsa Cabrera, John Calambokidis, Júlio Cardoso, Emma Carroll, Amina Cesario, Barbara Cheney, Ted Cheeseman, Enrico Corsi, Jens Currie, John Durban, Erin Falcone, Holly Fearnbach, Kiirsten Flynn, Trish Franklin, Dr Wally Franklin, Barbara Galletti Vernazzani, Tilen Genov, Marie Hill, David Johnston, Erin Keene, Claire Lacey, Sabre Mahaffy, Tamara McGuire, Liah McPherson, Catherine Meyer, Robert Michaud, Anastasia Miliou, Grace Olson, Dara Orbach, Heidi Pearson, Marianne Helene Rasmussen, Will Rayment, Renaldo Rinaldi, Salvatore Siciliano, Stephanie Stack, Beatriz Tintore, Leigh Torres, Jared Towers, Reny Tyson Moore, Caroline Weir, Rebecca Wellard, Randall Wells, Kym Yano, Jochen Zaeschmar and Lars Bejder.
Abstract: Several legal acts and international agreements mandate that scientific bodies regularly assess marine mammal stocks. Cetacean stock assessments often include abundance estimates from analyses of photographic identification (photo-ID) data, which involve processing large numbers of photos. To ease this burden, researchers increasingly make use of automated matching algorithms. These algorithms confront researchers with an opportunity, reducing the cost of population assessments, and a challenge, propagating photo-ID errors into abundance estimators at a large scale. In this study, we explore the tradeoffs between labor costs and estimator performance in population assessments relying on automated photo-ID. We used the real-life error rates from a multispecies matching algorithm, evaluated on 39 catalogs representing 24 species. We developed a general optimization tool for finding the optimal action amongst a set, which included true automation—where the model generated capture-histories without user input—to five degrees of partial automation—where users selected the correct match from 5, 10, …, 25 suggested matches. We found that true automation was optimal for catalogs where the algorithm matched images well. As matching performance declined, the tool tended to recommend that users evaluate more suggested matches from the algorithm, particularly for smaller catalogs. We found that false negative errors strongly predicted estimator performance, with a 2% increase in the false negative rate translating to a 5% increase in the relative bias in abundance. Researchers can use our tool to estimate expected performance of the abundance estimator, project labor effort, and find the optimal degree of automation for their catalog and algorithm. We recommend that researchers evaluate their algorithm’s performance, particularly by estimating the false negative rate, before deploying it in a population assessment. Our framework provides researchers with insight into the conservation benefits and consequences of automation as we enter a new era of artificial intelligence for stock assessments.
Workshop
Title: SMMILE Toolbox: Building Inclusive Science Education Opportunities
Organizers: Kirby Parnell and Brijonnay Madrigal
Abstract: Science education is an excellent way to communicate science efforts, inspire the next generation of scientists and/or engage with the local community. In an effort to bridge the gap between laboratory-based science efforts and the local community, the Marine Mammal Research Program developed the Summer Marine Mammal Intensive Learning Experience (SMMILE). SMMILE is a 10-day immersive program which provides high school students from Hawaiʻi and American Samoa a unique opportunity to learn about marine mammal research, conservation, and marine protected areas within the framework of traditional ecological knowledge (TEK) in Hawaiʻi. SMMILE was developed to increase representation of Native Hawaiians and Pacific Islanders in the field of marine mammal science. The purpose of this marine mammal education workshop will be to provide a "SMMILE toolbox" for scientists, professionals, and graduate students to implement programs in their community. During this workshop, we plan to discuss the following topics: curriculum development, incorporating TEK and place-based learning, funding, and advertising. Education program coordinators from around the world will be invited to present about their respective outreach and teaching opportunities to provide diverse examples of successful programs. Workshop coordinators and invited speakers will lead demonstrations of the hands-on activities included in their education programs. Ultimately, this workshop aims to provide attendees with opportunities to better integrate marine mammal outreach and educational opportunities within their communities.