MSBWO: A Multi-Strategies Improved Beluga Whale Optimization Algorithm for Feature Selection.

Feature selection (FS) is a classic and challenging optimization task in most machine learning and data mining projects. Recently, researchers have attempted to develop more effective methods by using metaheuristic methods in FS. To increase population diversity and further improve the effectiveness of the beluga whale optimization (BWO) algorithm, in this paper, we propose a multi-strategies improved BWO (MSBWO), which incorporates improved circle mapping and dynamic opposition-based learning (ICMDOBL) population initialization as well as elite pool (EP), step-adaptive Lévy flight and spiral updating position (SLFSUP), and golden sine algorithm (Gold-SA) strategies. Among them, ICMDOBL contributes to increasing the diversity during the search process and reducing the risk of falling into local optima. The EP technique also enhances the algorithm's ability to escape from local optima. The SLFSUP, which is distinguished from the original BWO, aims to increase the rigor and accuracy of the development of local spaces. Gold-SA is introduced to improve the quality of the solutions. The hybrid performance of MSBWO was evaluated comprehensively on IEEE CEC2005 test functions, including a qualitative analysis and comparisons with other conventional methods as well as state-of-the-art (SOTA) metaheuristic approaches that were introduced in 2024. The results demonstrate that MSBWO is superior to other algorithms in terms of accuracy and maintains a better balance between exploration and exploitation. Moreover, according to the proposed continuous MSBWO, the binary MSBWO variant (BMSBWO) and other binary optimizers obtained by the mapping function were evaluated on ten UCI datasets with a random forest (RF) classifier. Consequently, BMSBWO has proven very competitive in terms of classification precision and feature reduction.

Optimizing Deep Learning Models with Improved BWO for TEC Prediction.

The prediction of total ionospheric electron content (TEC) is of great significance for space weather monitoring and wireless communication. Recently, deep learning models have become increasingly popular in TEC prediction. However, these deep learning models usually contain a large number of hyperparameters. Finding the optimal hyperparameters (also known as hyperparameter optimization) is currently a great challenge, directly affecting the predictive performance of the deep learning models. The Beluga Whale Optimization (BWO) algorithm is a swarm intelligence optimization algorithm that can be used to optimize hyperparameters of deep learning models. However, it is easy to fall into local minima. This paper analyzed the drawbacks of BWO and proposed an improved BWO algorithm, named FAMBWO (Firefly Assisted Multi-strategy Beluga Whale Optimization). Our proposed FAMBWO was compared with 11 state-of-the-art swarm intelligence optimization algorithms on 30 benchmark functions, and the results showed that our improved algorithm had faster convergence speed and better solutions on almost all benchmark functions. Then we proposed an automated machine learning framework FAMBWO-MA-BiLSTM for TEC prediction, where MA-BiLSTM is for TEC prediction and FAMBWO for hyperparameters optimization. We compared it with grid search, random search, Bayesian optimization algorithm and beluga whale optimization algorithm. Results showed that the MA-BiLSTM model optimized by FAMBWO is significantly better than the MA-BiLSTM model optimized by grid search, random search, Bayesian optimization algorithm, and BWO.

Relationship between nutritional condition and causes of death in beluga whales Delphinapterus leucas from the St. Lawrence estuary, Quebec, Canada.

Carcasses of endangered beluga whales Delphinapterus leucas from the St. Lawrence Estuary, Canada, have been examined consistently since 1983 to determine causes of death. The objective of this study is to compare the nutritional condition of belugas that died of different causes. Previously published categories of death were refined to discriminate acute from chronic pathological processes. Bayesian linear models were used to predict cause of death from the scaled mass index (SMI). Causes of death were as follows: 'bacterial diseases', 'verminous pneumonia', 'toxoplasmosis', 'other parasitic diseases', 'other infectious diseases', 'trauma-entrapment', 'other noninfectious diseases', 'dystocia-postpartum complications', 'neonatal mortality', 'cancer', 'primary starvation' and 'undetermined'. The models predicted a lower nutritional condition for the 'neonatal mortality' in belugas <290 cm in length and for 'primary starvation' and 'verminous pneumonia' categories for belugas ≥290 cm. Belugas that died from 'dystocia-postpartum complications' or from 'undetermined causes' had a higher-than-average SMI. Animals in the 'trauma-entrapment' category did not exhibit the highest nutritional condition, which was unexpected since individuals that died from trauma or entrapment are often used as references for optimal nutritional condition in other cetacean populations. Females that died from dystocia and postpartum complications were in similar nutritional condition as females dead from other causes during, or shortly after, pregnancy. This suggests that these females are not obese, ruling out a possible cause of dystocia. Although studying dead animals biases results toward low nutritional condition, our findings support the link between chronic pathological processes and poorer nutritional condition in belugas.

Blueberry bruise non-destructive detection based on hyperspectral information fusion combined with multi-strategy improved Beluga Whale Optimization algorithm.

Introduction: Mechanical damage significantly reduces the market value of fruits, making the early detection of such damage a critical aspect of agricultural management. This study focuses on the early detection of mechanical damage in blueberries (variety: Sapphire) through a non-destructive method. Methods: The proposed method integrates hyperspectral image fusion with a multi-strategy improved support vector machine (SVM) model. Initially, spectral features and image features were extracted from the hyperspectral information using the successive projections algorithm (SPA) and Grey Level Co-occurrence Matrix (GLCM), respectively. Different models including SVM, RF (Random Forest), and PLS-DA (Partial Least Squares Discriminant Analysis) were developed based on the extracted features. To refine the SVM model, its hyperparameters were optimized using a multi-strategy improved Beluga Whale Optimization (BWO) algorithm. Results: The SVM model, upon optimization with the multi-strategy improved BWO algorithm, demonstrated superior performance, achieving the highest classification accuracy among the models tested. The optimized SVM model achieved a classification accuracy of 95.00% on the test set. Discussion: The integration of hyperspectral image information through feature fusion proved highly efficient for the early detection of bruising in blueberries. However, the effectiveness of this technology is contingent upon specific conditions in the detection environment, such as light intensity and temperature. The high accuracy of the optimized SVM model underscores its potential utility in post-harvest assessment of blueberries for early detection of bruising. Despite these promising results, further studies are needed to validate the model under varying environmental conditions and to explore its applicability to other fruit varieties.

Platelet phosphatidylserine exposure and microparticle production as health bioindicators in marine mammals.

In human medicine, various pathologies, including decompression sickness, thrombocytopenia, and rheumatoid arthritis, have been linked to changes in cellular microparticles (MP) formation, particularly platelet microparticles (PMP). Similar disorders in marine mammals might be attributed to anthropogenic threats or illnesses, potentially impacting blood PMP levels. Thus, detecting platelet phosphatidylserine (PS) exposure and PMP formation could serve as a crucial diagnostic and monitoring approach for these conditions in marine mammals. Our group has developed a methodology to assess real-time PS exposure and PMP formation specifically tailored for marine mammals. This method, pioneered in species such as bottlenose dolphins, beluga whales, walruses, and California sea lions, represents a novel approach with significant implications for both clinical assessment and further research into platelet function in these animals. The adapted methodology for evaluating PS exposure and PMP formation in marine mammals has yielded promising results. By applying this approach, we have observed significant correlations between alterations in PMP levels and specific pathologies or environmental factors. These findings underscore the potential of platelet function assessment as a diagnostic and monitoring tool in marine mammal health. The successful adaptation and application of this methodology in marine mammals highlight its utility for understanding and managing health concerns in these animals.

Intelligent and sustainable waste classification model based on multi-objective beluga whale optimization and deep learning.

Resource recycling is considered necessary for sustainable development, especially in smart cities where increased urbanization and the variety of waste generated require the development of automated waste management models. The development of smart technology offers a possible alternative to traditional waste management techniques that are proving insufficient to reduce the harmful effects of trash on the environment. This paper proposes an intelligent waste classification model to enhance the classification of waste materials, focusing on the critical aspect of waste classification. The proposed model leverages the InceptionV3 deep learning architecture, augmented by multi-objective beluga whale optimization (MBWO) for hyperparameter optimization. In MBWO, sensitivity and specificity evaluation criteria are integrated linearly as the objective function to find the optimal values of the dropout period, learning rate, and batch size. A benchmark dataset, namely TrashNet is adopted to verify the proposed model's performance. By strategically integrating MBWO, the model achieves a considerable increase in accuracy and efficiency in identifying waste materials, contributing to more effective waste management strategies while encouraging sustainable waste management practices. The proposed intelligent waste classification model outperformed the state-of-the-art models with an accuracy of 97.75%, specificity of 99.55%, F1-score of 97.58%, and sensitivity of 98.88%.

Research on Sea State Signal Recognition Based on Beluga Whale Optimization-Slope Entropy and One Dimensional-Convolutional Neural Network.

This study introduces a novel nonlinear dynamic analysis method, known as beluga whale optimization-slope entropy (BWO-SlEn), to address the challenge of recognizing sea state signals (SSSs) in complex marine environments. A method of underwater acoustic signal recognition based on BWO-SlEn and one-dimensional convolutional neural network (1D-CNN) is proposed. Firstly, particle swarm optimization-slope entropy (PSO-SlEn), BWO-SlEn, and Harris hawk optimization-slope entropy (HHO-SlEn) were used for feature extraction of noise signal and SSS. After 1D-CNN classification, BWO-SlEn were found to have the best recognition effect. Secondly, fuzzy entropy (FE), sample entropy (SE), permutation entropy (PE), and dispersion entropy (DE) were used to extract the signal features. After 1D-CNN classification, BWO-SlEn and 1D-CNN were found to have the highest recognition rate compared with them. Finally, compared with the other six recognition methods, the recognition rates of BWO-SlEn and 1D-CNN for the noise signal and SSS are at least 6% and 4.75% higher, respectively. Therefore, the BWO-SlEn and 1D-CNN recognition methods proposed in this paper are more effective in the application of SSS recognition.

Improved beluga whale optimization algorithm based cluster routing in wireless sensor networks.

Cluster routing is a critical routing approach in wireless sensor networks (WSNs). However, the uneven distribution of selected cluster head nodes and impractical data transmission paths can result in uneven depletion of network energy. For this purpose, we introduce a new routing strategy for clustered wireless sensor networks that utilizes an improved beluga whale optimization algorithm, called tCBWO-DPR. In the selection process of cluster heads, we introduce a new excitation function to evaluate and select more suitable candidate cluster heads by establishing the correlation between the energy of node and the positional relationship of nodes. In addition, the beluga whale optimization (BWO) algorithm has been improved by incorporating the cosine factor and t-distribution to enhance its local and global search capabilities, as well as to improve its convergence speed and ability. For the data transmission path, we use Prim's algorithm to construct a spanning tree and introduce DPR for determining the optimal route between cluster heads based on the correlation distances of cluster heads. This effectively shortens the data transmission path and enhances network stability. Simulation results show that the improved beluga whale optimization based algorithm can effectively improve the survival cycle and reduce the average energy consumption of the network.

Bronchial Tree System Analysis of Live Beluga Whale (Delphinapterus leucas) Using Bronchoscopy.

Cetaceans, including beluga whales (Delphinapterus leucas), have high morbidity and mortality rates due to bacterial or fungal lower respiratory infections. Bronchoalveolar lavage fluid (BALF) collection by bronchoscopy is beneficial for detecting pathogenic microorganisms in the lower respiratory tract. Efficient and safe bronchoscopy requires characterizing the bronchial tree systems of beluga whales, as no reports exist on bronchial length and bifurcation. In this study, bronchoscopy was performed on five captive beluga whales (9-44 years old) to detect bronchial length and bifurcation. The lengths from the blowhole to the scope impassable points due to the minimized bronchi diameters of the left principal bronchus (LPB), right principal bronchus (RPB), and tracheal bronchus (TB) were 110-155, 110-150, and 80-110 cm, respectively, and were correlated with the body length. Bronchoscopy identified more than 10, 10, and 6 bifurcated bronchi from the LPB, RPB, and TB, respectively. This is the first report to clarify the differences in bronchial tree systems between beluga whales and other cetaceans, as well as the differences for each individual beluga whale. These results could be useful for obtaining BALF via bronchoscopy to detect pathogenic microorganisms causing infections in the lower respiratory tract of beluga whales.

A Mine Water Source Prediction Model Based on LIF Technology and BWO-ELM.

The traditional methods for identifying water sources in coal mines lack the ability to quickly detect water sources and are prone to causing secondary pollution of samples. In contrast, laser induced fluorescence (LIF) technology has been introduced for the identification of coal mine water sources due to its high sensitivity and real-time performance. However, extreme learning machine (ELM) have shortcomings in randomly selecting weights and biases. The Beluga Whale Optimization (BWO) algorithm has efficient optimization capability, global search capability, adaptability and parallelism, and can find the optimal weights and biases in a short time. The combination of LIF technology and BWO-ELM model can be applied to quickly identify the welling water source in coal mine. Select sandstone water and old goaf water from the Huainan mining area as experimental samples, and mix them in different proportions to prepare 7 mixed water samples for testing. Utilize LIF technology to obtain spectral curve images, preprocess them with polynomial smoothing algorithm (SG) and spectral multiple scattering correction (MSC), and perform dimensionality reduction using factor analysis (FA) and linear discriminant analysis (LDA) methods. Finally, construct ELM models, Long Short Term Memory (LSTM) models, BWO-ELM models, and Particle Swarm Optimization Extreme Learning Machine(PSO-ELM) models for the dimensionality reduced data. In order to improve the reliability and accuracy of the results, the experimental results were kept to 5 decimal places. From the experimental results, it can be seen that SG-LDA-BWO-ELM has the best fitting effect, with a fitting coefficient of 0.99990, a root mean square error of 0.00041, a mean square error approaching 0, and an average absolute error of 0.00021. It has the best convergence and the smallest absolute error among all models, making it the most suitable for identifying mine water inrush. It is of great significance for preventing and controlling mine water disasters and ensuring coal mine production safety.

Midazolam and butorphanol combination for sedating beluga whales (Delphinapterus leucas).

Safe sedation doses for performing minor procedures such as bronchoscopy, endoscopy, and tooth extraction for beluga whales (Delphinapterus leucas) require elucidation. This study aimed to provide suggestions for determining appropriate midazolam and butorphanol doses to adequately sedate beluga whales to complete procedures and minimize the risk of side effects. We administered midazolam and butorphanol to six captive beluga whales (9-44 years old). Topical lidocaine anesthesia was administered during bronchoscopy. The sedation doses for the beluga whales varied from 0.020 to 0.122 mg/kg for midazolam and from 0.020 to 0.061 mg/kg for butorphanol. In beluga whales, optimal midazolam and butorphanol doses were lowest in old whales. These findings contribute to knowledge regarding appropriate sedation and prevention of overdose accidents during minor procedures in beluga whales.

Imitation of Novel Intransitive Body Actions in a Beluga Whale (Delphinapterus leucas): A "Do as Other Does" Study.

Cetaceans are well known for their unique behavioral habits, such as calls and tactics. The possibility that these are acquired through social learning continues to be explored. This study investigates the ability of a young beluga whale to imitate novel behaviors. Using a do-as-other-does paradigm, the subject observed the performance of a conspecific demonstrator involving familiar and novel behaviors. The subject: (1) learned a specific 'copy' command; (2) copied 100% of the demonstrator's familiar behaviors and accurately reproduced two out of three novel actions; (3) achieved full matches on the first trial for a subset of familiar behaviors; and (4) demonstrated proficiency in coping with each familiar behavior as well as the two novel behaviors. This study provides the first experimental evidence of a beluga whale's ability to imitate novel intransitive (non-object-oriented) body movements on command. These results contribute to our understanding of the remarkable ability of cetaceans, including dolphins, orcas, and now beluga whales, to engage in multimodal imitation involving sounds and movements. This ability, rarely documented in non-human animals, has significant implications for the development of survival strategies, such as the acquisition of knowledge about natal philopatry, migration routes, and traditional feeding areas, among these marine mammals.

Identification of two novel papillomaviruses in belugas.

Introduction: Papillomaviruses (PVs) can cause hyperplasia in the skin and mucous membranes of humans, mammals, and non-mammalian animals, and are a significant risk factor for cervical and genital cancers. Methods: Using next-generation sequencing (NGS), we identified two novel strains of papillomavirus, PV-HMU-1 and PV-HMU-2, in swabs taken from belugas (Delphinapterus leucas) at Polar Ocean Parks in Qingdao and Dalian. Results: We amplified the complete genomes of both strains and screened ten belugas and one false killer whale (Pseudorca crassidens) for the late gene (L1) to determine the infection rate. In Qingdao, 50% of the two sampled belugas were infected with PV-HMU-1, while the false killer whale was negative. In Dalian, 71% of the eight sampled belugas were infected with PV-HMU-2. In their L1 genes, PV-HMU-1 and PV-HMU-2 showed 64.99 and 68.12% amino acid identity, respectively, with other members of Papillomaviridae. Phylogenetic analysis of combinatorial amino acid sequences revealed that PV-HMU-1 and PV-HMU-2 clustered with other known dolphin PVs but formed distinct branches. PVs carried by belugas were proposed as novel species under Firstpapillomavirinae. Conclusion: The discovery of these two novel PVs enhances our understanding of the genetic diversity of papillomaviruses and their impact on the beluga population.

Do beluga whales truly migrate? Testing a key trait of the classical migration syndrome.

BACKGROUND: Migration enables organisms to access resources in separate regions that have predictable but asynchronous spatiotemporal variability in habitat quality. The classical migration syndrome is defined by key traits including directionally persistent long-distance movements during which maintenance activities are suppressed. But recently, seasonal round-trip movements have frequently been considered to constitute migration irrespective of the traits required to meet this movement type, conflating common outcomes with common traits required for a mechanistic understanding of long-distance movements. We aimed to test whether a cetacean ceases foraging during so-called migratory movements, conforming to a trait that defines classical migration. METHODS: We used location and dive data collected by satellite tags deployed on beluga whales (Delphinapterus leucas) from the Eastern Beaufort Sea population, which undertake long-distance directed movements between summer and winter areas. To identify phases of directionally persistent travel, behavioural states (area-restricted search, ARS; or Transit) were decoded using a hidden-Markov model, based on step length and turning angle. Established dive profiles were then used as a proxy for foraging, to test the hypothesis that belugas cease foraging during these long-distance transiting movements, i.e., they suppress maintenance activities. RESULTS: Belugas principally made directed horizontal movements when moving between summer and winter residency areas, remaining in a Transit state for an average of 75.4% (range = 58.5-87.2%) of the time. All individuals, however, exhibited persistent foraging during Transit movements (75.8% of hours decoded as the Transit state had ≥ 1 foraging dive). These data indicate that belugas actively search for and/or respond to resources during these long-distance movements that are typically called a migration. CONCLUSIONS: The long-distance movements of belugas do not conform to the traits defining the classical migration syndrome, but instead have characteristics of both migratory and nomadic behaviour, which may prove adaptive in the face of unpredictable environmental change. Such patterns are likely present in other cetaceans that have been labeled as migratory. Examination of not only horizontal movement state, but also the vertical behaviour of aquatic animals during directed movements is essential for identifying whether a species exhibits traits of the classical migration syndrome or another long-distance movement strategy, enabling improved ecological inference.

An improved multi-strategy beluga whale optimization for global optimization problems.

This paper presents an improved beluga whale optimization (IBWO) algorithm, which is mainly used to solve global optimization problems and engineering problems. This improvement is proposed to solve the imbalance between exploration and exploitation and to solve the problem of insufficient convergence accuracy and speed of beluga whale optimization (BWO). In IBWO, we use a new group action strategy (GAS), which replaces the exploration phase in BWO. It was inspired by the group hunting behavior of beluga whales in nature. The GAS keeps individual belugas whales together, allowing them to hide together from the threat posed by their natural enemy, the tiger shark. It also enables the exchange of location information between individual belugas whales to enhance the balance between local and global lookups. On this basis, the dynamic pinhole imaging strategy (DPIS) and quadratic interpolation strategy (QIS) are added to improve the global optimization ability and search rate of IBWO and maintain diversity. In a comparison experiment, the performance of the optimization algorithm (IBWO) was tested by using CEC2017 and CEC2020 benchmark functions of different dimensions. Performance was analyzed by observing experimental data, convergence curves, and box graphs, and the results were tested using the Wilcoxon rank sum test. The results show that IBWO has good optimization performance and robustness. Finally, the applicability of IBWO to practical engineering problems is verified by five engineering problems.

Skin Cortisol and Acoustic Activity: Potential Tools to Evaluate Stress and Welfare in Captive Cetaceans.

As people's focus broadens from animals on farms to zoos and aquaria, the field of welfare science and the public's concern for animal welfare continue to grow. In captive animals, stress and its causes are topics of interest in welfare issues, and the identification of an objective method that can be used to assess animals' stress as a physiological state is essential. Both behavioral and physiological parameters can be used as indicators in order to assess animal stress quantitatively. To validate this approach, acoustic activity and the sloughed scrape skin cortisol concentration were used to evaluate the animal welfare of captive beluga whales (Delphinapterus leucas). The acoustic activity (5 min at 10:00 am) of three captive D. leucas was routinely recorded by a transducer and analyzed using audio editing software. The calls were separated into three main categories: whistles, pulses, and combo calls. The sloughed scrape skin samples were collected non-invasively once a week from all three animals' fluke and/or flipper. Cortisol was extracted using a modified skin steroid extraction technique, and detected via commercially available enzyme immunoassays. The results showed that the cortisol concentration increased by varying levels when the whales encountered the same event. In addition, the number and distribution of the calls changed along with the events. This indicated that the changes in the cortisol concentration and acoustic behavior may have reflected the fluctuations in the environment and body condition. Therefore, the scrape cortisol measurement and acoustic recordings could be used to monitor stress levels in captive beluga whales. We recommend that aquaria consider incorporating skin scrape cortisol and acoustic activity monitoring into their standards for animal welfare.

Crystal Structures of Inhibitor-Bound Main Protease from Delta- and Gamma-Coronaviruses.

With the spread of SARS-CoV-2 throughout the globe causing the COVID-19 pandemic, the threat of zoonotic transmissions of coronaviruses (CoV) has become even more evident. As human infections have been caused by alpha- and beta-CoVs, structural characterization and inhibitor design mostly focused on these two genera. However, viruses from the delta and gamma genera also infect mammals and pose a potential zoonotic transmission threat. Here, we determined the inhibitor-bound crystal structures of the main protease (Mpro) from the delta-CoV porcine HKU15 and gamma-CoV SW1 from the beluga whale. A comparison with the apo structure of SW1 Mpro, which is also presented here, enabled the identification of structural arrangements upon inhibitor binding at the active site. The cocrystal structures reveal binding modes and interactions of two covalent inhibitors, PF-00835231 (active form of lufotrelvir) bound to HKU15, and GC376 bound to SW1 Mpro. These structures may be leveraged to target diverse coronaviruses and toward the structure-based design of pan-CoV inhibitors.

Declining concentrations of chlorinated paraffins in endangered St. Lawrence Estuary belugas (Delphinapterus leucas): Response to regulations or a change in diet?

Very high levels of industrial contaminants in St. Lawrence Estuary (SLE) beluga whales represent one of the major threats to this population classified as endangered under the Species at Risk Act in Canada. Elevated concentrations of short-chained chlorinated paraffins (SCCPs) were recently reported in blubber of adult male SLE belugas. Recent regulations for SCCPs in North America, combined with their replacement by medium- (MCCPs) and long-chained chlorinated paraffins (LCCPs), highlight the importance of tracking this toxic chemical class. The objectives of this study were to evaluate (1) levels and profiles of chlorinated paraffins (CPs) in samples obtained from carcasses of adult male, adult female, juvenile, newborn, and fetus beluga, and (2) trends in adult male belugas between 1997 and 2018. Factors potentially influencing CP temporal trends such as age, feeding ecology and sampling year were also explored. SCCPs dominated (64 to 100%) total CP concentrations across all age and sex classes, MCCPs accounted for the remaining proportion of total CPs, and LCCPs were not detected in any sample. The chlorinated paraffin homolog that dominated the most in beluga blubber was C12Cl8. Adult male SCCP concentrations from this study were considerably lower (> 2000-fold) than those recently reported in Simond et al. (2020), likely reflecting a previously erroneous overestimate due to the lack of a suitable analytical method for SCCPs at the time. Both SCCPs and total CPs declined over time in adult males in our study (rate of 1.67 and 1.33% per year, respectively), presumably due in part to the implementation of regulations in 2012. However, there is a need to better understand the possible contribution of a changing diet to contaminant exposure, as stable isotopic ratios of carbon also changed over time.

Combining δ13C and δ15N from bone and dentine in marine mammal palaeoecological research: insights from toothed whales.

Stable carbon (δ13C) and nitrogen (δ15N) isotopic compositions of bone and dentine collagen extracted from museum specimens have been widely used to study the paleoecology of past populations. Due to possible systematic differences in stable isotope values between bone and dentine, dentine values need to be transformed into bone-collagen equivalent using a correction factor to allow comparisons between the two collagen sources. Here, we provide correction factors to transform dentine δ13C and δ15N values into bone-collagen equivalent for two toothed whales: narwhal and beluga. We sampled bone and dentine from the skulls of 11 narwhals and 26 belugas. In narwhals, dentine was sampled from tusk and embedded tooth; in belugas, dentine was sampled from tooth. δ13C and δ15N were measured, and intra-individual bone and dentine isotopic compositions were used to calculate correction factors for each species. We detected differences in δ13C and δ15N. In both narwhals and belugas, we found lower average δ13C and δ15N in bone compared with dentine. The correction factors provided by the study enable the combined analysis of stable isotope data from bone and dentine in these species.

Machine-Learning Approach for Automatic Detection of Wild Beluga Whales from Hand-Held Camera Pictures.

A key aspect of ocean protection consists in estimating the abundance of marine mammal population density within their habitat, which is usually accomplished using visual inspection and cameras from line-transect ships, small boats, and aircraft. However, marine mammal observation through vessel surveys requires significant workforce resources, including for the post-processing of pictures, and is further challenged due to animal bodies being partially hidden underwater, small-scale object size, occlusion among objects, and distracter objects (e.g., waves, sun glare, etc.). To relieve the human expert's workload while improving the observation accuracy, we propose a novel system for automating the detection of beluga whales (Delphinapterus leucas) in the wild from pictures. Our system relies on a dataset named Beluga-5k, containing more than 5.5 thousand pictures of belugas. First, to improve the dataset's annotation, we have designed a semi-manual strategy for annotating candidates in images with single (i.e., one beluga) and multiple (i.e., two or more belugas) candidate subjects efficiently. Second, we have studied the performance of three off-the-shelf object-detection algorithms, namely, Mask-RCNN, SSD, and YOLO v3-Tiny, on the Beluga-5k dataset. Afterward, we have set YOLO v3-Tiny as the detector, integrating single- and multiple-individual images into the model training. Our fine-tuned CNN-backbone detector trained with semi-manual annotations is able to detect belugas despite the presence of distracter objects with high accuracy (i.e., 97.05 mAP@0.5). Finally, our proposed method is able to detect overlapped/occluded multiple individuals in images (beluga whales that swim in groups). For instance, it is able to detect 688 out of 706 belugas encountered in 200 multiple images, achieving 98.29% precision and 99.14% recall.