Search

Abstract Machine learning‐based behaviour classification using acceleration data is a powerful tool in bio‐logging research. Deep learning architectures such as convolutional neural networks (CNN), long short‐term memory (LSTM) and self‐attention mechanism as well as related training techniques have been extensively studied in human activity recognition. However, they have rarely been used in wild animal studies. The main challenges of acceleration‐based wild animal behaviour classification include data shortages, class imbalance problems, various types of noise in data due to differences in individual behaviour and where the loggers were attached and complexity in data due to complex animal‐specific behaviours, which may have limited the application of deep learning techniques in this area. To overcome these challenges, we explored the effectiveness of techniques for efficient model training: data augmentation, manifold mixup and pre‐training of deep learning models with unlabelled data, using datasets from two species of wild seabirds and state‐of‐the‐art deep learning model architectures. Data augmentation improved the overall model performance when one of the various techniques (none, scaling, jittering, permutation, time‐warping and rotation) was randomly applied to each data during mini‐batch training. Manifold mixup also improved model performance, but not as much as random data augmentation. Pre‐training with unlabelled data did not improve model performance. The state‐of‐the‐art deep learning models, including a model consisting of four CNN layers, an LSTM layer and a multi‐head attention layer, as well as its modified version with shortcut connection, showed better performance among other comparative models. Using only raw acceleration data as inputs, these models outperformed classic machine learning approaches that used 119 handcrafted features. Our experiments showed that deep learning techniques are promising for acceleration‐based behaviour classification of wild animals and highlighted some challenges (e.g. effective use of unlabelled data). There is scope for greater exploration of deep learning techniques in wild animal studies (e.g. advanced data augmentation, multimodal sensor data use, transfer learning and self‐supervised learning). We hope that this study will stimulate the development of deep learning techniques for wild animal behaviour classification using time‐series sensor data.

Abstract Research on predator–prey interactions between sharks and cetaceans remain limited. Here, we report on a video of a shortfin mako shark circling a finless porpoise with a damaged caudal fin in the Seto Inland Sea, Japan. The finless porpoise was neither emaciated nor inactive, but unable to swim effectively due to the complete lack of a caudal fin. Some circumstantial evidence, including a bite mark on the porpoise's head, strongly suggests that the mako shark attacked it. Furthermore, the possible time difference between the two injuries the porpoise sustained may reflect the shark's hunting tactics. While mako sharks primarily feed on small fish and cephalopods, this observation suggests they also may prey on live cetaceans more often than previously thought.

Abstract Plastic pollution is distributed patchily around the world’s oceans. Likewise, marine organisms that are vulnerable to plastic ingestion or entanglement have uneven distributions. Understanding where wildlife encounters plastic is crucial for targeting research and mitigation. Oceanic seabirds, particularly petrels, frequently ingest plastic, are highly threatened, and cover vast distances during foraging and migration. However, the spatial overlap between petrels and plastics is poorly understood. Here we combine marine plastic density estimates with individual movement data for 7137 birds of 77 petrel species to estimate relative exposure risk. We identify high exposure risk areas in the Mediterranean and Black seas, and the northeast Pacific, northwest Pacific, South Atlantic and southwest Indian oceans. Plastic exposure risk varies greatly among species and populations, and between breeding and non-breeding seasons. Exposure risk is disproportionately high for Threatened species. Outside the Mediterranean and Black seas, exposure risk is highest in the high seas and Exclusive Economic Zones (EEZs) of the USA, Japan, and the UK. Birds generally had higher plastic exposure risk outside the EEZ of the country where they breed. We identify conservation and research priorities, and highlight that international collaboration is key to addressing the impacts of marine plastic on wide-ranging species.

Abstract Background Understanding what animals do in time and space is important for a range of ecological questions, however accurate estimates of how animals use space is challenging. Within the use of animal-attached tags, radio telemetry (including the Global Positioning System, ‘GPS’) is typically used to verify an animal’s location periodically. Straight lines are typically drawn between these ‘Verified Positions’ (‘VPs’) so the interpolation of space-use is limited by the temporal and spatial resolution of the system’s measurement. As such, parameters such as route-taken and distance travelled can be poorly represented when using VP systems alone. Dead-reckoning has been suggested as a technique to improve the accuracy and resolution of reconstructed movement paths, whilst maximising battery life of VP systems. This typically involves deriving travel vectors from motion sensor systems and periodically correcting path dimensions for drift with simultaneously deployed VP systems. How often paths should be corrected for drift, however, has remained unclear. Methods and results Here, we review the utility of dead-reckoning across four contrasting model species using different forms of locomotion (the African lion Panthera leo, the red-tailed tropicbird Phaethon rubricauda, the Magellanic penguin Spheniscus magellanicus, and the imperial cormorant Leucocarbo atriceps). Simulations were performed to examine the extent of dead-reckoning error, relative to VPs, as a function of Verified Position correction (VP correction) rate and the effect of this on estimates of distance moved. Dead-reckoning error was greatest for animals travelling within air and water. We demonstrate how sources of measurement error can arise within VP-corrected dead-reckoned tracks and propose advancements to this procedure to maximise dead-reckoning accuracy. Conclusions We review the utility of VP-corrected dead-reckoning according to movement type and consider a range of ecological questions that would benefit from dead-reckoning, primarily concerning animal–barrier interactions and foraging strategies.

Abstract Background Fine-scale data on animal position are increasingly enabling us to understand the details of animal movement ecology and dead-reckoning, a technique integrating motion sensor-derived information on heading and speed, can be used to reconstruct fine-scale movement paths at sub-second resolution, irrespective of the environment. On its own however, the dead-reckoning process is prone to cumulative errors, so that position estimates quickly become uncoupled from true location. Periodic ground-truthing with aligned location data (e.g., from global positioning technology) can correct for this drift between Verified Positions (VPs). We present step-by-step instructions for implementing Verified Position Correction (VPC) dead-reckoning in R using the tilt-compensated compass method, accompanied by the mathematical protocols underlying the code and improvements and extensions of this technique to reduce the trade-off between VPC rate and dead-reckoning accuracy. These protocols are all built into a user-friendly, fully annotated VPC dead-reckoning R function; Gundog.Tracks, with multi-functionality to reconstruct animal movement paths across terrestrial, aquatic, and aerial systems, provided within the Additional file 4 as well as online (GitHub). Results The Gundog.Tracks function is demonstrated on three contrasting model species (the African lion Panthera leo, the Magellanic penguin Spheniscus magellanicus, and the Imperial cormorant Leucocarbo atriceps) moving on land, in water and in air. We show the effect of uncorrected errors in speed estimations, heading inaccuracies and infrequent VPC rate and demonstrate how these issues can be addressed. Conclusions The function provided will allow anyone familiar with R to dead-reckon animal tracks readily and accurately, as the key complex issues are dealt with by Gundog.Tracks. This will help the community to consider and implement a valuable, but often overlooked method of reconstructing high-resolution animal movement paths across diverse species and systems without requiring a bespoke application.

Abstract Background Recent advances in sensing technologies have enabled us to attach small loggers to animals in their natural habitat. It allows measurement of the animals’ behavior, along with associated environmental and physiological data and to unravel the adaptive significance of the behavior. However, because animal-borne loggers can now record multi-dimensional (here defined as multimodal) time series information from a variety of sensors, it is becoming increasingly difficult to identify biologically important patterns hidden in the high-dimensional long-term data. In particular, it is important to identify co-occurrences of several behavioral modes recorded by different sensors in order to understand an internal hidden state of an animal because the observed behavioral modes are reflected by the hidden state. This study proposed a method for automatically detecting co-occurrence of behavioral modes that differs between two groups (e.g., males vs. females) from multimodal time-series sensor data. The proposed method first extracted behavioral modes from time-series data (e.g., resting and cruising modes in GPS trajectories or relaxed and stressed modes in heart rates) and then identified two different behavioral modes that were frequently co-occur (e.g., co-occurrence of the cruising mode and relaxed mode). Finally, behavioral modes that differ between the two groups in terms of the frequency of co-occurrence were identified. Results We demonstrated the effectiveness of our method using animal-locomotion data collected from male and female Streaked Shearwaters by showing co-occurrences of locomotion modes and diving behavior recorded by GPS and water-depth sensors. For example, we found that the behavioral mode of high-speed locomotion and that of multiple dives into the sea were highly correlated in male seabirds. In addition, compared to the naive method, the proposed method reduced the computation costs by about 99.9%. Conclusion Because our method can automatically mine meaningful behavioral modes from multimodal time-series data, it can be potentially applied to analyzing co-occurrences of locomotion modes and behavioral modes from various environmental and physiological data.

Abstract Anthropogenic marine debris is a threat to marine organisms. Understanding how this debris spatially distributes at sea and may become associated with marine wildlife are key steps to tackle this current issue. Using bird-borne GPS- and video-loggers on 13 black-footed albatrosses Phoebastria nigripes breeding in Torishima, Japan, we examined the distribution of large floating debris in the Kuroshio Current area, western North Pacific. A total of 16 floating debris, including styrofoam (n = 4), plastic pieces (n = 3), plastic sheet (n = 1), fishery-related items (rope or netting, n = 4), and unidentified debris (n = 4), were recorded across the 9003 km covered by nine birds. The debris was concentrated in the southern area of the Kuroshio Current, where the surface current was weak, and the albatrosses were foraging. The albatrosses displayed changes in flight direction towards the debris when at a mean distance of 4.9 km, similarly to when approaching prey, and one bird was observed pecking at a plastic sheet; indicating that albatrosses actively interacted with the debris. This paper shows the usefulness of studying wide-ranging marine predators through the use of combined biologging tools, and highlights areas with increased risk of debris exposure and behavioral responses to debris items.

Abstract Background Salmonids return to the river where they were born in a phenomenon known as mother-river migration. The underpinning of migration has been extensively examined, particularly regarding the behavioral correlations of external environmental cues such as the scent of the mother-river and geomagnetic compass. However, neuronal underpinning remains elusive, as there have been no biologging techniques suited to monitor neuronal activity in the brain of large free-swimming fish. In this study, we developed a wireless biologging system to record extracellular neuronal activity in the brains of free-swimming salmonids. Results Using this system, we recorded multiple neuronal activities from the telencephalon of trout swimming in a rectangular water tank. As proof of principle, we examined the activity statistics for extracellular spike waveforms and timing. We found cells firing maximally in response to a specific head direction, similar to the head direction cells found in the rodent brain. The results of our study suggest that the recorded signals originate from neurons. Conclusions We anticipate that our biologging system will facilitate a more detailed investigation into the neural underpinning of fish movement using internally generated information, including responses to external cues.

Comparative analysis of animal behaviour using locomotion data such as GPS data is difficult because the large amount of data makes it difficult to contrast group differences. Here the authors apply deep learning to detect and highlight trajectories characteristic of a group across scales of millimetres to hundreds of kilometres.

Animal behavior is the final and integrated output of brain activity. Thus, recording and analyzing behavior is critical to understand the underlying brain function. While recording animal behavior has become easier than ever with the development of compact and inexpensive devices, detailed behavioral data analysis requires sufficient prior knowledge and/or high content data such as video images of animal postures, which makes it difficult for most of the animal behavioral data to be efficiently analyzed. Here, we report a versatile method using a hybrid supervised/unsupervised machine learning approach for behavioral state estimation and feature extraction (STEFTR) only from low-content animal trajectory data. To demonstrate the effectiveness of the proposed method, we analyzed trajectory data of worms, fruit flies, rats, and bats in the laboratories, and penguins and flying seabirds in the wild, which were recorded with various methods and span a wide range of spatiotemporal scales—from mm to 1,000 km in space and from sub-seconds to days in time. We successfully estimated several states during behavior and comprehensively extracted characteristic features from a behavioral state and/or a specific experimental condition. Physiological and genetic experiments in worms revealed that the extracted behavioral features reflected specific neural or gene activities. Thus, our method provides a versatile and unbiased way to extract behavioral features from simple trajectory data to understand brain function.

Telomeres are nucleotide sequences located at the ends of chromosomes that promote genome stability. Changes in telomere length (dynamics) are related to fitness or life expectancy, and telomere dynamics during the development phase are likely to be affected by growth and stress factors. Here, we examined telomere dynamics of black-tailed gull chicks (Larus crassirostris) in nests with and without siblings. We found that the initial telomere lengths of singletons at hatching were longer than those of siblings, indicating that singletons are higher-quality chicks than siblings in terms of telomere length. Other factors likely affecting individual quality (i.e., sex, laying date, laying order of eggs, and clutch size) were not related to telomere lengths. Within broods, initial telomere lengths were longer in older chicks than in younger chicks, suggesting that maternal effects, which vary with laying sequence, influence the initial lengths. Additionally, telomeres of chicks with a sibling showed more attrition between hatching and fledging than those of singleton chicks, suggesting that being raised with siblings can cause a sustained competitive environment that leads to telomere loss. High growth rates were associated with a low degree of telomere shortening observed in older siblings, perhaps because slower growth reflects higher food stress and/or higher aerobic metabolism from increased begging effort. Our results show that developmental telomere attrition was an inevitable consequence in two-chick nests in the pre- and post-hatching microenvironments due to the combination of social stress within the nest and maternal effects. The results of our study shed light on telomere dynamics in early life, which may represent an important physiological undercurrent of life-history traits.

Summary The doubly labelled water (DLW) method is an isotope-based technique that is used to measure the metabolic rates of free-living animals. We validated the DLW method for measuring metabolic rates in five rhinoceros auklets (Cerorhinca monocerata) compared with simultaneous measurements using the respirometric method. We calculated the CO2 production rate of four auklets (mean initial body mass: 552 g±36 s.d.) injected with DLW, using the one- and two-pool models. The metabolic rate during the 24-h measurements in a respirometric chamber for resting auklets averaged 16.30±1.66 kJ h−1 (n = 4). The metabolic rates determined using the one- and two-pool models in the DLW method for the same period as the respirometric measurement averaged 16.61±2.13 kJ h−1 (n = 4) and 16.16±2.10 kJ h−1 (n = 4), respectively. The mean absolute percent error between the DLW and respirometric methods was 8.04% using the one-pool model and was slightly better than that with the two-pool model. The differences in value between the DLW and respirometric methods are probably due to oxygen isotope turnover, which eliminated only 10–14% of the initial enrichment excess.

During the last few years, the development of animal-borne still cameras and video recorders has enabled researchers to observe what a wild animal sees in the field. In the present study, we deployed miniaturized video recorders to investigate the underwater foraging behavior of Imperial cormorants (Phalacrocorax atriceps). Video footage was obtained from 12 animals and 49 dives comprising a total of 8.1 h of foraging data. Video information revealed that Imperial cormorants are almost exclusively benthic feeders. While foraging along the seafloor, animals did not necessarily keep their body horizontal but inclined it downwards. The head of the instrumented animal was always visible in the videos and in the majority of the dives it was moved constantly forward and backward by extending and contracting the neck while travelling on the seafloor. Animals detected prey at very short distances, performed quick capture attempts and spent the majority of their time on the seafloor searching for prey. Cormorants foraged at three different sea bottom habitats and the way in which they searched for food differed between habitats. Dives were frequently performed under low luminosity levels suggesting that cormorants would locate prey with other sensory systems in addition to sight. Our video data support the idea that Imperial cormorants' efficient hunting involves the use of specialized foraging techniques to compensate for their poor underwater vision.

The mesopelagic zone of the northeast Pacific Ocean is an important foraging habitat for many predators, yet few studies have addressed the factors driving basin-scale predator distributions or inter-annual variability in foraging and breeding success. Understanding these processes is critical to reveal how conditions at sea cascade to population-level effects. To begin addressing these challenging questions, we collected diving, tracking, foraging success, and natality data for 297 adult female northern elephant seal migrations from 2004 to 2010. During the longer post-molting migration, individual energy gain rates were significant predictors of pregnancy. At sea, seals focused their foraging effort along a narrow band corresponding to the boundary between the sub-arctic and sub-tropical gyres. In contrast to shallow-diving predators, elephant seals target the gyre-gyre boundary throughout the year rather than follow the southward winter migration of surface features, such as the Transition Zone Chlorophyll Front. We also assessed the impact of added transit costs by studying seals at a colony near the southern extent of the species' range, 1,150 km to the south. A much larger proportion of seals foraged locally, implying plasticity in foraging strategies and possibly prey type. While these findings are derived from a single species, the results may provide insight to the foraging patterns of many other meso-pelagic predators in the northeast Pacific Ocean.

In seabirds, the relationship between flight speed and wind direction/speed is thought to be particularly important for studying energy-saving strategy and foraging habitat selection. In this study, we examined whether the ground and calculated air speeds of four great cormorants (Phalacrocorax carbo) were affected by wind conditions using high-resolution GPS data loggers. The birds increased their ground flight speed in tailwinds, decreased it in headwinds, and changed their air speed in relation to wind components. However, they did not change their foraging sites according to the wind conditions. They were likely to respond to moderate wind conditions by adjusting their air speed without changing their foraging sites.

To analyze an animal's movement trajectory, a basic model is required that satisfies the following conditions: the model must have an ecological basis and the parameters used in the model must have ecological interpretations, a broad range of movement patterns can be explained by that model, and equations and probability distributions in the model should be mathematically tractable. Random walk models used in previous studies do not necessarily satisfy these requirements, partly because movement trajectories are often more oriented or tortuous than expected from the models. By improving the modeling for turning angles, this study aims to propose a basic movement model. On the basis of the recently developed circular auto-regressive model, we introduced a new movement model and extended its applicability to capture the asymmetric effects of external factors such as wind. The model was applied to GPS trajectories of a seabird (Calonectris leucomelas) to demonstrate its applicability to various movement patterns and to explain how the model parameters are ecologically interpreted under a general conceptual framework for movement ecology. Although it is based on a simple extension of a generalized linear model to circular variables, the proposed model enables us to evaluate the effects of external factors on movement separately from the animal's internal state. For example, maximum likelihood estimates and model selection suggested that in one homing flight section, the seabird intended to fly toward the island, but misjudged its navigation and was driven off-course by strong winds, while in the subsequent flight section, the seabird reset the focal direction, navigated the flight under strong wind conditions, and succeeded in approaching the island.

Most seabirds are diurnal foragers, but some species may also feed at night. In Peruvian pelicans (Pelecanus thagus), the evidence for nocturnal foraging is sparse and anecdotal. We used GPS-dataloggers on five incubating Peruvian pelicans from Isla Lobos de Tierra, Perú, to examine their nocturnality, foraging movements and activities patterns at sea. All instrumented pelicans undertook nocturnal trips during a 5-7 day tracking period. Eighty-seven percent of these trips (n = 13) were strictly nocturnal, whereas the remaining occurred during the day and night. Most birds departed from the island after sunset and returned a few hours after sunrise. Birds traveled south of the island for single-day trips at a maximum range of 82.8 km. Overall, 22% of the tracking period was spent at sea, whereas the remaining time was spent on the island. In the intermediate section of the trip (between inbound and outbound commutes), birds spent 77% of the trip time in floating bouts interspersed by short flying bouts, the former being on average three times longer than the latter. Taken together, the high sinuosity of the bird's tracks during floating bouts, the exclusively nocturnal trips of most individuals, and the fact that all birds returned to the island within a few hours after sunrise suggest that pelicans were actively feeding at night. The nocturnal foraging strategy of Peruvian pelicans may reduce food competition with the sympatric and strictly diurnal Guanay cormorants (Phalacrocorax bougainvillii), Peruvian boobies (Sula variegata) and Blue-footed boobies (S. nebouxii), which were present on the island in large numbers. Likewise, plankton bioluminescence might be used by pelicans as indirect cues to locate anchovies during their upward migration at night. The foraging success of pelicans at night may be enhanced by seizing prey close to the sea surface using a sit-and-wait strategy.

While social interactions play a crucial role on the development of young individuals, those of highly mobile juvenile birds in inaccessible environments are difficult to observe. In this study, we deployed miniaturised video recorders on juvenile brown boobies Sula leucogaster, which had been hand-fed beginning a few days after hatching, to examine how social interactions between tagged juveniles and other birds affected their flight and foraging behaviour. Juveniles flew longer with congeners, especially with adult birds, than solitarily. In addition, approximately 40% of foraging occurred close to aggregations of congeners and other species. Young seabirds voluntarily followed other birds, which may directly enhance their foraging success and improve foraging and flying skills during their developmental stage, or both.

Animal movement represents an interdisciplinary research field involving ecology and physics. Animal locomotion is composed of three interrelated hierarchies: micro-scale mechanisms of locomotion, often in the realm of physics; intermediate-scale regulation of speed and direction and macro-scale movement of route selection, those can significantly affect the fitness of the animal. Based on these hierarchies, animal movement strategies can be formulated as optimisation problems, yielding some theoretical predictions that can now be tested using bio-logging techniques to record animal movements. This review focuses on bird movement strategies in response to wind and aims to provide a practical overview of how theory and empirical testing with bio-logging data are carried out in the field of movement ecology. [ABSTRACT FROM AUTHOR]