Fig 4 - uploaded by Jorge Fontes
Content may be subject to copyright.
Classification tree and confusion matrix. (A) The learned decision tree classifier. A two-dimensional vector containing the maximum ODBA peak and average power spectral density (PSD) in a 1 s window is input into the tree. The tree makes a series of branching linear separations, visualized as a set of branching points, and produces the gait output seen at the leaves of the tree. (B) The accuracy of the automatic detection algorithms as visualized using a confusion matrix. The three main gaits identified were finning, flapping and jetting. Sixty-four video clips (maximum and minimum durations of 73 s and 4 s, respectively; mean duration of 13 s) of tagged squid gait events were analyzed by a human operator and these results were compared with the automatic classification algorithm. The rows represent the human-labeled gaits, and the columns represent labels generated by the automatic classifier. The number in each cell corresponds to the proportion of gait events classified with a specific combination of true and predicted label. For reference, a perfect classifier would show only detections along the diagonal squares because true and predicted labels would always correspond.
Source publication
Squid are mobile, diverse, ecologically important marine organisms whose behavior and habitat use can have substantial impacts on ecosystems and fisheries. However, due in part to the inherent challenges of monitoring squid in their natural marine environment, fine-scale behavioral observations of these free-swimming, soft-bodied animals are rare....
Contexts in source publication
Context 1
... to match the fundamental frequency of dynamic accelerations of the slowest observed gait present in the dataset, i.e. finning, which was an approximately 1 Hz signal. Our analysis used the peak ODBA magnitude and the average power spectral density in each window to classify gait using a decision tree classifier (Kotsiantis, 2013) (see Fig. 3 and Fig. 4A). These features were chosen because preliminary analysis of the data showed that the dynamic acceleration signal exhibited different magnitudes and frequencies for the three gaits. Finning was correlated with low ODBA magnitude and a low spectral density concentrated near 1 Hz; flapping was correlated with higher ODBA and spectral ...
Context 2
... across the dataset ceased to be sensitive to changes in window size, following the method in Shepard et al. (2008). We found this point to be approximately 50 samples, or 2 s of data at 25 Hz. classification and the model parameters were optimized to maximize classifier accuracy on the training dataset. The resulting decision tree is shown in Fig. 4A, while the ODBA acceleration and spectral features are shown in Fig. 3 for a 15 s segment from the training ...
Context 3
... videos of the squid were analyzed and each gait event with the tagged squid in the video frame was given a gait label. The performance of the gait classifier compared with the human-labeled behaviors was visualized in a confusion matrix (Fig. 4B). The overall accuracy of the gait classifier was 99.6% (595 of 597 s classified correctly). The algorithm had high classification accuracy for finning and jetting gaits (100%: 573 of 573 correct; and 100%: 6 of 6 correct, respectively); flapping detection was slightly less accurate (89% or 16 of 18 correct). All misclassifications for ...
Citations
... In this sense, our research is among the first studies correctly address trait comparisons for marine animals, suggesting that further research should incorporate this approach in macroecological studies. New approaches to the knowledge of the distributional range of mobile species, such as the bio-logging tags (Flaspohler et al. 2019;Cones et al. 2022), or eDNA could help in accurate assessments of the real extent of the species, as well as their biological activity and behavioural patterns. ...
Macroecological studies have primarily focused on investigating the relationships between body size and geographic distribution on large scales, including regional, continental, and even global levels. While the majority of these studies have been conducted on terrestrial species, a limited number of studies have been carried out on aquatic species, and even fewer have considered the importance of phylogeny in the observed patterns. Cephalopods provide a good model for examining these macroecological patterns due to their large geographic and bathymetric ranges, wide range of body sizes, as well as diverse fin sizes and shapes. In this study, we assess the relationships between mantle length, fin size, and hatchling size with the geographic and bathymetric distribution of 30 squid species from the worldwide distributed family Loliginidae. To test a macroecological hypothesis, we evaluated the phylogenetic signal and correlated evolution to assess the role of biological traits in squid distribution, using a molecular phylogeny based on two mitochondrial and one nuclear genes. Biological traits (mantle length and fin size) exhibit high phylogenetic signals, while distribution demonstrates low signal. The correlation analyses revealed the existence of a relationship between adult mantle length and fin size with geographic and bathymetric distribution, but not with hatchling size. The geographic distribution of loliginid squids evolved in relation to mantle length, where larger squids with large fins (e.g. Sepioteuthis) have wide distributions, while small-finned species (e.g. Pickfordia-teuthis) have narrow distributions. This study paves the way for exploring similar relationships in other squid families or other marine swimming animals.
... ITAGs were used to measure squid movement dynamics. The sensor package was small (length: 7 cm, width: 3 cm, height: 1 cm) and was affixed using surgical sutures (Mooney et al., 2015;Flaspohler et al., 2019;Cones et al., 2022). Additionally, ITAGs were neutrally buoyant, hydrodynamic, and focal tagged squid exhibited normal swimming and schooling behaviors with other conspecifics. ...
... Intense jet propulsion events are high acceleration movements employed in response to predators or during conspecific interactions, but is also the common response of squid to recorded pile driving noise (Wells and O'Dor, 1991;Hanlon et al., 2002;Jones et al., 2020). The jetting gait was quantified using similar methods described in detail in previous studies (Flaspohler et al., 2019;Cones et al., 2022). In brevity, a movement was deemed a jetting event if ODBA exceeded 0.3 gravities (g). ...
Anthropogenic noise is now a prominent pollutant increasing in both terrestrial and marine environments. In the ocean, proliferating offshore windfarms, a key renewable energy source, are a prominent noise concern, as their pile driving construction is among the most intense anthropogenic sound sources. Yet, across taxa, there is little information of pile driving noise impacts on organismal fine-scale movement despite its key link to individual fitness. Here, we experimentally quantified the swimming behavior of an abundant squid species (Doryteuthis pealeii) of vital commercial and ecological importance in response to in situ pile driving activity on multiple temporal and spatial scales (thus exposed to differing received levels, or noise-doses). Pile driving induced energetically costly alarm-jetting behaviors in most (69%) individuals at received sound levels (in zero to peak) of 112-123 dB re 1 µm s⁻², levels similar to those measured at the kilometer scale from some wind farm construction areas. No responses were found at a comparison site with lower received sound levels. Persistence of swimming pattern changes during noise-induced alarm responses, a key metric addressing energetic effects, lasted up to 14 s and were significantly shorter in duration than similar movement changes caused by natural conspecific interactions. Despite observing dramatic behavioral changes in response to initial pile driving noise, there was no evidence of gait changes over an experiment day. These results demonstrate that pile driving disrupts squid fine-scale movements, but impacts are short-lived suggesting that offshore windfarm construction may minimally impact the energetics of this ecologically key taxon. However, further work is needed to assess potential behavioral and physiological impacts at higher noise levels.
... In this sense, our research is among the rst studies correctly addressing trait comparisons for marine animals, suggesting that further research should incorporate this approach in macroecological studies. New approaches to the knowledge of the distributional range of mobile species, such as the bio-logging tags (Flaspohler et al. 2019;Cones et al. 2022) could help in an accurate determination of the real extent of the species, as well as their biological activity and behavioural patterns. ...
Macroecological studies have mainly focused on exploring the relationships between body size and geographic distribution on large scales, whether regional, continental or even global, and most of them have been conducted on terrestrial species. Few studies have been conducted on aquatic species, and even fewer have considered the importance of phylogeny in the observed patterns. Cephalopod molluscs are a good model to tackle these problems given that they have large geographic and bathymetric ranges, a wide range of body sizes, as well as diverse fin sizes and shapes. Here, we evaluate the relationships between body and fin size with the geographic distribution of 30 squid species of the family Loliginidae distributed worldwide. To test a macroecological hypothesis, we evaluated the phylogenetic signal and correlated evolution of the three traits to assess the role of phylogenetic relationships in squid distribution using a molecular phylogeny based on two mitochondrial and one nuclear gene. The analyses showed the existence of a relationship between body size and geographic distribution. Similarly, relative fin size showed a positive relationship with distribution. Phylogenetic signals were high for morphological traits (body and fin size), while it was low for distribution. The geographic distribution of loliginid squids evolved in relation to body size, where larger squids with large fins (e.g. genus Sepioteuthis ) have wide distributions, while small-finned species (e.g. genus Pickfordioteuthis ) have narrow distributions. This study opens the gates to explore such relationships in other squid families or other marine swimming animals.
... In addition, marine animal telemetry through tagging is also an important emerging field benefiting from new technologies and automatisation of data processing, with crucial implications for animal conservation and management (Allen and Singh, 2016;Hays et al., 2016). Bio-logging technology has improved over the past several years and tags are increasingly able to measure a range of movements, behaviors and physiological parameters of different marine organisms (Flaspohler et al., 2019;Savoca et al., 2021), with also environmental data for days or weeks at a time (Martıń Loṕez et al., 2022). For example, AniBOS (Animal Borne Ocean Sensors) is an emerging network of the Global Ocean Observing System (GOOS), as a project of the OD program "Ocean Observing Co-Design -Evolving ocean observing for a sustainable future" 5 . ...
The Ocean Decade (OD) is the name of a United Nations (UN) initiative devoted to ocean science for sustainable development. It started in 2021 and will provide an opportunity to create a new foundation, across the science-policy interface, to strengthen the sustainable management of oceans and coasts and, at same time, bring benefits for future generations. The OD will establish a common strategy to achieve the objectives of the 2030 Agenda for Sustainable Development and the other European Union (EU) and international agreements, such as the EU Marine Strategy Framework Directive (MSFD), the Marine Biodiversity Strategy and the UN World Assessment I and II aimed at preserving ocean health. Furthermore, the OD has several expected outcomes that concern different aspects of marine environment, including its enormous values for humans. Several entities will combine efforts to achieve these goals and science is at the forefront of the sustainable blue economy. Marine research is solving complex challenges through interdisciplinary approaches, revolutionizing our life and our interaction with the ocean. This review discusses recent advancements in science related to the OD outcomes. The role of new technology for ocean exploration and monitoring, the importance of omics science and biotechnology to deal with ocean pollution, and other innovative solutions are discussed. All of these are inspired by the idea of using marine resources in a sustainable way and without impacting in a negative way on marine ecosystems. The role of science communication is therefore considered a crucial issue to spread the OD messages and to reach the general public and stakeholders. Only by the integration of science, governance, industry and public, the OD will have any chance to succeed.
... Unlike other jet-propelling taxa, squid can propel by a combination of fin and jet propulsion. During low-speed swimming, both fin and jet propulsion are used, with 'finning' gaits used most frequently, especially when ascending from vertical migrations (Flaspohler et al., 2019). As swimming speed increases, jet propulsion increasingly contributes to total thrust production among shallow-water squid species. ...
Pulsatile jet propulsion is a common swimming mode used by a diverse array of aquatic taxa from chordates to cnidarians. This mode of locomotion has interested both biologists and engineers for over a century. A central issue to understanding the important features of jet-propelling animals is to determine how the animal interacts with the surrounding fluid. Much of our knowledge of aquatic jet propulsion has come from simple theoretical approximations of both propulsive and resistive forces. Although these models and basic kinematic measurements have contributed greatly, they alone cannot provide the detailed information needed for a comprehensive, mechanistic overview of how jet propulsion functions across multiple taxa, size scales and through development. However, more recently, novel experimental tools such as high-speed 2D and 3D particle image velocimetry have permitted detailed quantification of the fluid dynamics of aquatic jet propulsion. Here, we provide a comparative analysis of a variety of parameters such as efficiency, kinematics and jet parameters, and review how they can aid our understanding of the principles of aquatic jet propulsion. Research on disparate taxa allows comparison of the similarities and differences between them and contributes to a more robust understanding of aquatic jet propulsion.
... Bio-logging of marine invertebrates, such as cephalopods, crustaceans and cnidarians, was found to be scarce (so far 17 publications, which represented only about 3.4% of all studies) (Fig. 2). Soft-bodied eco-sensor tags had been applied for trial attachments to jellyfish (Aurelia aurita), squid (Loligo forbesi), veined squids (Loligo forbesii), and Pacific sea nettles (Chrysaora fuscescens) and these studies presented good examples of estimating swimming movements of invertebrates with vertical and horizontal movement (Mooney et al. 2015;Fannjiang et al. 2019;Flaspohler et al. 2019). Although there has been remarkable progress in bio-logging, bio-logging on invertebrates are relatively poorly represented even though invertebrates are widely and diversely distributed living organisms in the ocean. ...
Recent technologies have allowed researchers to observe animal behaviour and monitor their surrounding environments by deploying electronic sensors onto the animals. So-called ‘bio-logging’ (also known as animal telemetry, biotelemetry, or animal-borne sensors) has been widely used to study marine animals that are difficult for humans to observe. In this study, we (1) review the types of sensors used, the animal taxa studied, and the study areas in marine bio-logging publications from 1974 to 2019; (2) introduce the main topics in behavioural and environmental marine bio-logging studies; and (3) discuss suggestions for future marine bio-logging studies. We expect that technological advances in new sensors will enhance the ability of both behavioural ecologists and oceanographers to explore animal movements, physiology and marine environments. In addition, we discuss future perspectives of bio-loggers to improve data acquisition and accuracy with longer battery life for applying bio-logging techniques to broader species.
Marine animals equipped with sensors provide vital information for understanding their ecophysiology and collect oceanographic data on climate change and for resource management. Existing methods for attaching sensors to marine animals mostly rely on invasive physical anchors, suction cups, and rigid glues. These methods can suffer from limitations, particularly for adhering to soft fragile marine species such as squid and jellyfish, including slow complex operations, unreliable fixation, tissue trauma, and behavior changes of the animals. However, soft fragile marine species constitute a significant portion of ocean biomass (>38.3 teragrams of carbon) and global commercial fisheries. Here we introduce a soft hydrogel-based bioadhesive interface for marine sensors that can provide rapid (time <22 s), robust (interfacial toughness >160 J m⁻²), and non-invasive adhesion on various marine animals. Reliable and rapid adhesion enables large-scale, multi-animal sensor deployments to study biomechanics, collective behaviors, interspecific interactions, and concurrent multi-species activity. These findings provide a promising method to expand a burgeoning research field of marine bio-sensing from large marine mammals and fishes to small, soft, and fragile marine animals.
Diel vertical migration (DVM) is a vital behavior for many pelagic marine fauna. Locomotory tactics that animals use during DVM define the metabolic costs of migrations and influence the risk of detection and capture by predators, yet, for squids, there is little understanding of the fine-scale movements and potential variability during these migrations. Vertical migratory behaviors of 5 veined squid Loligo forbesii were investigated with biologging tags (ITags) off the Azores Islands (central North Atlantic). Diel movements ranged from 400 to 5 m and were aligned with sunset and sunrise. During ascent periods, 2 squid exhibited cyclic climb-and-glide movements using primarily jet propulsion, while 3 squid ascended more continuously and at a lower vertical speed using mostly a finning gait. Descents for all 5 squid were consistently more rapid and direct. While all squid swam in both arms-first and mantle-first directions during DVM, mantle-first swimming was more common during upward movements, particularly at vertical speeds greater than 25 cm s ⁻¹ . The in situ variability of animal posture, swim direction, and gait use revealed behavioral flexibility interpreted as energy conservation, prey capture, and predator avoidance.
Background
Studies on animal behaviour often involve the quantification of the occurrence and duration of various activities. When direct observations are challenging (e.g., at night, in a burrow, at sea), animal-borne devices can be used to remotely record the movement and behaviour of an animal (e.g., changing body posture and movement, geographical position) and/or its immediate surrounding environment (e.g., wet or dry, pressure, temperature, light). Changes in these recorded variables are related to different activities undertaken by the animal. Here we explored the use of animal-borne acoustic recorders to automatically infer activities in seabirds.
Results
We deployed acoustic recorders on Cape gannets and analysed sound data from 10 foraging trips. The different activities (flying, floating on water and diving) were associated with clearly distinguishable acoustic features. We developed a method to automatically identify the activities of equipped individuals, exclusively from animal-borne acoustic data. A random subset of four foraging trips was manually labelled and used to train a classification algorithm (k-nearest neighbour model). The algorithm correctly classified activities with a global accuracy of 98.46%. The model was then used to automatically assess the activity budgets on the remaining non-labelled data, as an illustrative example. In addition, we conducted a systematic review of studies that have previously used data from animal-borne devices to automatically classify animal behaviour ( n = 61 classifications from 54 articles). The majority of studies (82%) used accelerometers (alone or in combination with other sensors, such as gyroscopes or magnetometers) for classifying activities, and to a lesser extent GPS, acoustic recorders or pressure sensors, all potentially providing a good accuracy of classification (> 90%).
Conclusion
This article demonstrates that acoustic data alone can be used to reconstruct activity budgets with very good accuracy. In addition to the animal’s activity, acoustic devices record the environment of equipped animals (biophony, geophony, anthropophony) that can be essential to contextualise the behaviour of animals. They hence provide a valuable alternative to the set of tools available to assess animals’ behaviours and activities in the wild.
