Animal Behavior Research Articles

Monitoring poultry social dynamics using colored tags: Avian visual perception, behavioral effects, and artificial intelligence precision

Artificial intelligence (AI) in animal behavior and welfare research is on the rise. AI can detect behaviors and localize animals in video recordings, thus it is a valuable tool for studying social dynamics. However, maintaining the identity of individuals over time, especially in homogeneous poultry flocks, remains challenging for algorithms. We propose using differentially colored “backpack” tags (black, gray, white, orange, red, purple, and green) detectable with computer vision (eg. YOLO) from top-view video recordings of pens. These tags can also accommodate sensors, such as accelerometers. In separate experiments, we aim to: (i) evaluate avian visual perception of the different colored tags; (ii) assess the potential impact of tag colors on social behavior; and (iii) test the ability of the YOLO model to accurately distinguish between different colored tags on Japanese quail in social group settings. First, the reflectance spectra of tags and feathers were measured. An avian visual model was applied to calculate the quantum catches for each spectrum. Green and purple tags showed significant chromatic contrast to the feather. Mostly tags presented greater luminance receptor stimulation than feathers. Birds wearing white, gray, purple, and green tags pecked significantly more at their own tags than those with black (control) tags. Additionally, fewer aggressive interactions were observed in groups with orange tags compared to groups with other colors, except for red. Next, heterogeneous groups of 5 birds with different color tags were videorecorded for 1 h. The precision and accuracy of YOLO to detect each color tag were assessed, yielding values of 95.9% and 97.3%, respectively, with most errors stemming from misclassifications between black and gray tags. Lastly using the YOLO output, we estimated each bird's average social distance, locomotion speed, and the percentage of time spent moving. No behavioral differences associated with tag color were detected. In conclusion, carefully selected colored backpack tags can be identified using AI models and can also hold other sensors, making them powerful tools for behavioral and welfare studies.

The recurrent temporal restricted Boltzmann machine captures neural assembly dynamics in whole-brain activity

Animal behaviour alternates between stochastic exploration and goal-directed actions, which are generated by the underlying neural dynamics. Previously, we demonstrated that the compositional Restricted Boltzmann Machine (cRBM) can decompose whole-brain activity of larval zebrafish data at the neural level into a small number (∼100-200) of assemblies that can account for the stochasticity of the neural activity (van der Plas et al., eLife, 2023). Here, we advance this representation by extending to a combined stochastic-dynamical representation to account for both aspects using the recurrent temporal RBM (RTRBM) and transfer-learning based on the cRBM estimate. We demonstrate that the functional advantage of the RTRBM is captured in the temporal weights on the hidden units, representing neural assemblies, for both simulated and experimental data. Our results show that the temporal expansion outperforms the stochastic-only cRBM in terms of generalization error and achieves a more accurate representation of the moments in time. Lastly, we demonstrate that we can identify the original time-scale of assembly dynamics by estimating multiple RTRBMs at different temporal resolutions. Together, we propose that RTRBMs are a valuable tool for capturing the combined stochastic and time-predictive dynamics of large-scale data sets.

From data on gross activity to the characterization of animal behaviour: which metrics for which purposes?

From data on gross activity to the characterization of animal behaviour: which metrics for which purposes?

YOLOv5DA: An Improved YOLOv5 Model for Posture Detection of Grouped Pigs

Accurate posture detection is the foundation for analyzing animal behavior, which can promote animal welfare. With the development of computer vision, such technology has been widely used in analyzing animal behavior without physical contact. However, computer vision technology for pig posture detection often suffers from problems of missed or false detection due to complex scenarios. To solve the problem, this study proposed a novel object detection model YOLOv5DA, which was based on YOLOv5s and designed for pig posture detection from 2D camera video. Firstly, we established the annotated dataset (7220 images) including the training set (5776 images), validation set (722 images), and test set (722 images). Secondly, an object detection model YOLOv5DA based on YOLOv5s was proposed to recognize pig postures (standing, prone lying, and side lying), which incorporated Mosaic9 data augmentation, deformable convolution, and adaptive spatial feature fusion. The comparative and ablation experiments were conducted to verify the model’s effectiveness and reliability. Finally, we used YOLOv5DA to detect the posture distribution of pigs. The results revealed that the standing posture was more frequent in the morning and afternoon and the side-lying posture was most common at noon. This observation demonstrated that the posture of pigs is influenced by temperature variations. The study demonstrated that YOLOv5DA could accurately identify three postures of standing, prone lying, and side lying with an average precision (AP) of 99.4%, 99.1%, and 99.1%, respectively. Compared with YOLOv5s, YOLOv5DA could effectively handle occlusion while increasing the mean precision (mAP) by 1.7%. Overall, our work provided a highly accurate, effective, low-cost, and non-contact strategy of posture detection in grouped pigs, which can be used to monitor pig behavior and assist in the early prevention of disease.

The recurrent temporal restricted Boltzmann machine captures neural assembly dynamics in whole-brain activity.

Animal behaviour alternates between stochastic exploration and goal-directed actions, which are generated by the underlying neural dynamics. Previously, we demonstrated that the compositional Restricted Boltzmann Machine (cRBM) can decompose whole-brain activity of larval zebrafish data at the neural level into a small number (∼100-200) of assemblies that can account for the stochasticity of the neural activity (van der Plas et al., eLife, 2023). Here, we advance this representation by extending to a combined stochastic-dynamical representation to account for both aspects using the recurrent temporal RBM (RTRBM) and transfer-learning based on the cRBM estimate. We demonstrate that the functional advantage of the RTRBM is captured in the temporal weights on the hidden units, representing neural assemblies, for both simulated and experimental data. Our results show that the temporal expansion outperforms the stochastic-only cRBM in terms of generalization error and achieves a more accurate representation of the moments in time. Lastly, we demonstrate that we can identify the original time-scale of assembly dynamics by estimating multiple RTRBMs at different temporal resolutions. Together, we propose that RTRBMs are a valuable tool for capturing the combined stochastic and time-predictive dynamics of large-scale data sets.

Modelling behavior of Crested gecko (Correlophus ciliatus) using classification algorithms

Animal behavior plays a crucial role in evolution of many species. Many studies focused on animal behavior enhance the ability to collect large and detailed data. However, this kind of data is surpassing the capability of traditional statistical methods for analysis. In this study we propose to use artificial intelligence (AI) with machine learning models (ML) as tools to study animal behavior and potentially assumed evolution patterns in their behavior. For the Crested gecko (Correlophus ciliatus), some guidelines have been published regarding the breeding of these reptiles, focusing on their behavior. However, little is known about moderating their behavior using AI and advanced ML algorithms. In this study, based on information collected from twenty individuals, we proposed building a supervised classifier model using simple Decision Tree classifier (DT), Gradient Boosting classifier (GB) and Extreme Gradient Boosting classifier (XGBoost). Our results show that the highest accuracy (above 60 %) was achieved for variables which were not complex in terms of animal behavior. The analysis presented in this study, demonstrates that it is possible to model Crested Gecko behavior using ML models.

Onboard Sensors Reveal New Insights into Animal Decision-Making

The continuous process of decision-making in animals is crucial for their survival. For example, when deciding when, where, and with whom to forage, they need to consider their internal state, previous experience, and social information in addition to external factors such as food distribution and weather conditions. Studying animal decision-making in the wild is a complicated task due to the complexity of the process, which requires continuous monitoring of the examined individual and its environment. Here, we review the most advanced methods to examine decision-making from an individual point of view, namely tracking technologies to monitor the movement of an individual, the sensory information available to it, the presence and behavior of other animals around it, and its surrounding environment. We provide examples for studying decision-making during competition, examining the ontogeny of decision-making, and describing the importance of long-term monitoring and field manipulation for understanding decision processes throughout different life stages.

Targeting glycolysis for neuroprotection in early LPS-induced neuroinflammation

Neuroinflammation is a pathophysiological feature of numerous neurological and psychiatric disorders. The immune response in the central nervous system, driven by microglia and astrocytes, leads to metabolic reprogramming towards aerobic glycolysis, a phenomenon known as the Warburg effect. The control of metabolic reprogramming via immunomodulation may represent a potential therapeutic target for providing protection against neuroinflammation, which contributes to neuronal dysfunction and death in several neurological disorders. For this purpose, we investigated putative neuroprotective effects of the downregulation of aerobic glycolysis using the 3PO inhibitor, and the downregulation of neuroinflammation using MCC950, in the early LPS-induced neuroinflammation model. The LPS-induced shift towards glycolysis, inflammatory and glial changes (IL-1β, NF-κB, COX2, Iba1, GFAP) were reversed by 3PO, which improved animal behavior. Additionally, MCC950 (an NLRP3 inhibitor) downregulated TLR4/Akt/p38 MAPK/NF-κB/STAT3 signaling, expressions of COX2 and IL-1β, and the astrocyte reactivity (decreasing GFAP) induced by early neuroinflammation, resulting in low glucose uptake. Our data support the occurrence of the Warburg effect during early neuroinflammation and suggest potential new approaches for the treatment of brain injury, given the role of neuroinflammation in such events.

Animal Linguistics

Animal linguistics is an interdisciplinary field that integrates animal behavior, linguistics, and cognitive science to explore issues such as (a) what animal signals mean, (b) what cognitive abilities are necessary for the production and understanding of these signals, and (c) how communication systems have evolved. Despite the traditional belief that language evolved through a single mutation in our ancestors, accumulating evidence suggests that many cognitive abilities underlying human language have also evolved in nonhuman animals. For example, several species of birds and nonhuman primates convey conceptual meanings through specific vocalizations and/or combine multiple meaning-bearing calls into sequences using syntactic rules. Using experimental paradigms inspired by cognitive science and linguistics, animal linguistics aims to uncover the cognitive mechanisms underlying animal language and explores its evolutionary principles. This review examines previous studies exploring the meanings and cognitive abilities underlying animal language and introduces key methodologies in this emerging field.

Mitigation measures for pinniped-fisheries interactions based on knowledge of animal behavior

Abstract A rise in pinniped-fisheries interactions has been observed due to the global decline in fish populations primarily attributed to overfishing, and/or due to the recent population recovery of several pinniped species. Although many studies have quantified these interactions, a limited number delve into their behavioral aspects. We consulted 374 studies on pinniped-fisheries interactions from 1980 to 2020 and analyzed which of them addressed the interaction from a behavioral perspective. Additionally, we examined the role of learning in pinnipeds and their ability to capture prey from fishing gear. The stimulus-reward relationship in operational interactions was addressed in 51 (14%) studies, involving 10 (19%) of the world’s 54 pinniped species or subspecies. Depredation behavior (43%) and attack behavior (35%) were frequently reported during fishing interactions. Understanding associative and non-associative learning mechanisms has the potential to reduce economic losses from fishing activity, advance efficient mitigation measures to minimize conflicts, and establish pragmatic conservation priorities based on a better understanding of the behavioral aspects of these interactions. More research is needed to explore the effectiveness of deterrents in different species, improve methods of reducing habituation, and evaluate long-term behavioral effects, fostering effective strategies for minimizing conflicts and promoting coexistence between humans and pinnipeds.

PRÁTICAS DE ENSINO EM SALA DE AULA: IMPACTO DAS ATIVIDADES INTERATIVAS NO APRENDIZADO DE BIOLOGIA

This article investigates the impact of interactive activities proposed in the classroom on the biology learning of integrated high school students at the IFMT Várzea Grande campus. The methodology involved direct and indirect participant observation, with the aim of describing the behavior, feeding methods, reproduction and interactions of animals observed in the backyard of these students' homes. The results indicated an increase in engagement and understanding of the concepts studied in the classroom and that the backyard environment proved to be its own small ecosystem, adapted to the conditions and species that live there.

Loss of DOT1L function disrupts neuronal transcription, animal behavior, and leads to a novel neurodevelopmental disorder.

Individuals with monoallelic pathogenic variants in the histone lysine methyltransferase DOT1L display global developmental delay and varying congenital anomalies. However, the impact of monoallelic loss of DOT1L remains unclear. Here, we present a largely female cohort of 11 individuals with DOT1L variants with developmental delays and dysmorphic facial features. We found that DOT1L variants include missense variants clustered in the catalytic domain, frameshift, and stop-gain variants. We demonstrate that specific variants cause loss of methyltransferase activity and therefore sought to define the effects of decreased DOT1L function. Using RNA-sequencing of cultured neurons and single nucleus RNA-sequencing of mouse cortical tissue, we found that partial Dot1l depletion causes sex-specific transcriptional responses and disrupts transcription of synaptic genes. Further, Dot1l loss alters neuron branching and expression of synaptic proteins. Lastly using zebrafish and mouse models, we found behavioral disruptions that include sex-specific deficits in mice. Overall, we define how DOT1L loss leads to neurological dysfunction by demonstrating that partial Dot1l loss impacts transcription, neuron morphology, and behavior across multiple models and systems.

Protective Effects of Lemna minor Linn. On Hepatic and Cognitive Impairments in Acetaminophen Induced Hepatic Encephalopathy in Rats

Aim: Lemna minor Linn., an aquatic plant, is a promising novel therapeutic agent that has been traditionally used in ethnobotanical practices as an ecofriendly supplement for the management of various ailments. This study involves the evaluation of ethanolic extract of Lemna minor Linn. against Paracetamol-induced Hepatic Encephalopathy using in vitro and in vivo Models. Methods: The acute oral toxicity study of the ethanolic extract of Lemna minor (EELM) was conducted following the OECD-425 guidelines over a 14-day period. A total of nine animals were used for this toxicity assessment. The EELM was tested in Paracetamol(PCM)-induced bioactivation animal model at two different dosages 200 and in comparison with silymarin as a standard compound. The In vivo experimental study was conducted using Sprague Dawley rats which was divided into 5 groups each group containing 6 animals so the total no of animals used in the study was 30 animals. The treatment groups included: normal control ( ), 100mg/kg silymarin (standard) and the EELM of two different doses of 200 and , p.o were administered to rats 10 hr before paracetamol ( ) treatment. Rats were orally administered their respective doses every day for total 30days. Paracetamol-induced oxidative liver damage disrupted normal levels of liver enzymes, total protein, and bilirubin, while also depleting antioxidant reserves. This oxidative stress was strongly associated with paracetamol toxicity, leading to a marked depletion of glutathione (GSH) and impairing both memory and cognitive function in the animals. Behavioral parameters are performed to evaluate the effect of drugs on cognitive behaviour of animals. Morris water maze test was performed to study how animals learn and remembers the spatial information relying on distal cues to locate the hidden platform in an opaque water. Additionally, elevated plus maze was performed to measures the anxious behaviour of rats, the criterion was tested based on the conflict between rats innate instincts to explore new environment and avoid open, well-lit areas. The potential protective effects of EELM was evaluated by measuring serum enzyme levels and antioxidants status in the liver and brain, further histopathological analysis was performed respectively. Results: The acute toxicity study did not report any mortality or toxicity signs in animals. PCM toxicity led to a statistically increase in the liver and body weight, along with brain water content. The PCM-intoxicated group exhibited a marked reduction in the level of superoxide dismutase (SOD) and glutathione (GSH) in the brain and liver, as well as an increase in lipid peroxidation and serum biomarkers (AST, ALT, ALP, and total bilirubin). EELM significantly ( ) reduced liver injury by inhibiting ALT, AST and ALP levels in serum. SOD, GSH and MDA liver content were significantly ( ) elevated by EELM, compared to PCM treated rats. Comparing the treatment and induced group, the treated group successfully recovered the activity of antioxidant levels and also been acknowledged for restored liver functioning by alleviating oxidative stress and also GSH level in brain was significantly increased by EELM and preserved the histology of brain, which was chronically produced over a period of 30 days. Conclusion: The findings of this investigation indicate the traditional use of Lemna minor in hepatoprotection and neuroprotection by regulating oxidative stress and mitigating reactive oxygen species (ROS). The insights gained from this research contribute to the development of novel therapeutic agents and paves the way for further studies on Lemna minor to enhance health outcomes, particularly in the management of neurodegenerative diseases.

The study of the toxicological profile of dry extracts of narrow-leaved lavender herb

Aim. To study the acute toxicity of narrow-leaved lavender (Lavandula angustifolia) herb dry extracts of Ukrainian origin. Materials and methods. The study objects were experimental test samples: test sample No. 1 ‒ a dry extract of narrow-leaved lavender herb obtained by extraction with purified water; test sample No. 2 ‒ a dry extract of narrow-leaved lavender herb obtained by extraction with a water‒ethanol solution (40 % ethanol); test sample No. 3 ‒ a dry extract of narrow-leaved lavender herb obtained by extraction with a water‒ethanol solution (70 % ethanol). To reproduce the clinic of acute poisoning and determine LD50, the acute toxicity of test samples was studied on sexually mature white mice weighing 19-21 g under the conditions of a single intragastric administration. The animals were observed for 2 weeks, which made it possible to assess the toxic effect of the lavender extracts studied on the body of experimental animals. The weight (initial data, 3, 7 and 14 days), the behavior of animals after the introduction of test samples (appearance, respiration, salivation, urination, excreta, feed consumption by animals) were monitored by daily feed and water consumption ‒ on days 7 and 14. The macroscopic examination included the external examination of animals, the examination of internal organs of the chest (heart, lungs, thymus) and abdominal cavities (liver, spleen, kidneys, adrenal glands, testes/ovaries). For statistical processing of the data obtained (body weight), the arithmetic average value and its standard error were used; the reliability of discrepancies between samples was assessed using ANOVA RM and the Dunnet criterion. Comparisons were made with the negative control group. Indicators of the mass coefficients of internal organs were presented in the form of the median, upper and lower quartiles. The reliability of discrepancies between samples was assessed using the Kruskal-Wallis test and the Mann-Whitney test compared to the negative control group. Results and discussion. It was experimentally proven that intragastric administration of dry extracts from lavender herb in the dose of 5000 mg/kg did not lead to lethality and did not affect the general trophic processes of animals. The macroscopic examination of internal organs and analysis of their mass coefficients confirmed that a single intragastric administration of test samples of lavender extracts did not lead to pathological changes in both males and females. The organ mass coefficients were within the normal range and had no statistically significant differences compared to animals in the negative control group. Conclusions. The results of the study have demonstrated that a single intragastric administration of test samples of lavender in the dose of 5000 mg/kg does not lead to pathological changes in the physiological state of mice, which allows to attribute the extracts studied to class V of practically non-toxic substances (LD50 > 5000 mg/kg).

“Neuroinflammation”: does it have a role in chronic pain? Evidence from human imaging

Abstract Despite hundreds of studies demonstrating the involvement of neuron-glia-immune interactions in the establishment and/or maintenance of persistent pain behaviors in animals, the role (or even occurrence) of so-called “neuroinflammation” in human pain has been an object of contention for decades. Here, I present the results of multiple positron emission tomography (PET) studies measuring the levels of the 18 kDa translocator protein (TSPO), a putative neuroimmune marker, in individuals with various pain conditions. Overall, these studies suggest that brain TSPO PET signal: (1) is elevated, compared to healthy volunteers, in individuals with chronic low back pain (with additional elevations in spinal cord and neuroforamina), fibromyalgia, migraine and other conditions characterized by persistent pain; (2) has a spatial distribution exhibiting a degree of disorder specificity; (3) is parametrically linked to pain characteristics or comorbid symptoms (eg, nociplastic pain, fatigue, depression), as well as measures of brain function (ie, functional connectivity), in a regionally-specific manner. In this narrative, I also discuss important caveats to consider in the interpretation of this work (eg, regarding the cellular source of the signal and the complexities inherent in its acquisition and analysis). While the biological and clinical significance of these findings awaits further work, this emerging preclinical literature supports a role of neuron-glia-immune interactions as possible pathophysiological underpinnings of human chronic pain. Gaining a deeper understanding of the role of neuroimmune function in human pain would likely have important practical implications, possibly paving the way for novel interventions.

Animal emotions and consciousness: a preliminary assessment of researchers' perceptions and biases and prospects for future progress.

Scientists and philosophers have long struggled with the question of whether non-human animals experience emotions or consciousness. Yet, it is unclear where the scientific consensus on these topics lies today. To address this gap, we administered a survey of professional animal behaviour researchers to assess perceptions regarding (i) the taxonomic distribution of emotions and consciousness in non-human animals, (ii) respondents' confidence in this assessment, and (iii) attitudes towards pitfalls and potential for progress when addressing these questions. Respondents (n = 100) ascribe emotionality and consciousness to a broad swath of the animal taxonomy, including non-human primates, other mammals, birds and cephalopods. Respondents' attribution of these phenomena was strongly associated with their confidence in their assessments (R 2 > 0.9), with respondents assuming an absence of emotions and consciousness when they were unsure. We also identify an emergent consensus of the components involved in a functional definition of emotions. Researchers are optimistic that tools either currently exist or will exist in the future to rigorously address these questions (>85%) and that animal behaviour, as a field, should do more to encourage research works on emotions (>70%). We discuss implications for publication bias and future work in this area as well as ethical considerations regarding animal care and use.

Classifying active and inactive states of growing rabbits from accelerometer data using machine learning algorithms

Using wearable accelerometers is gaining traction in research and animal production management for monitoring animal behaviour. In this study, the objective was to automatically detect rabbit activity/inactivity states from accelerometer data in growing rabbits. For that purpose, 16 animals were equipped with an accelerometer and filmed for 2 weeks. A total of 10 hours of video across all the rabbits were annotated manually using the Boris software, identifying 6 classes of different behaviours: lying, eating, moving, grooming, walking and drinking which were grouped into two classes: active and inactive. Accelerometer signal and video annotations were manually synchronized. The static and dynamic components of the signal were isolated by applying a low-pass and high-pass filter and 4 additional time series were derived from these components. The signal was segmented into time windows of different sizes: 1, 3, 5, 7 and 9 seconds. For each window, a total of 41 features were extracted in the time and frequency domain. Different subsets of data containing an increasing number (from 5 to 25 in steps of 5) of the most informative features identified with random forest (RF) were used to train a binary classification model (inactive as a positive class). The classification performance of RF, support vector machine (SVM) and gradient boosting (GB) was evaluated. A nested cross-validation (CV) with an outer Leave-One-Animal Out CV and an inner threefold CV for hyperparameter tunning was implemented. The same resampling was implemented for each window size and each classifier so that the models were evaluated with the same data sets. The performance was evaluated on the test datasets using different metrics: precision, recall, F1 score and accuracy. Results showed that the classifiers perform very similarly. With the best configuration (window size of 9 s and with the 5 most important features) the RF model reaches a median precision of 1 (Q1=0.99, Q3=1) and a median recall of 0.93 (Q1=0.89, Q3=0.97). These results showed that the model is highly reliable in correctly classifying positive instances. Additionally, achieving a recall of 0.93 emphasizes the model's effectiveness in capturing a substantial portion of positive instances. Accelerometers combined with machine learning models therefore hold great promise for monitoring rabbit activity and for a range of applications in animal science and behaviours.

General anesthesia activates a central anxiolytic center in the BNST

Low doses of general anesthetics like ketamine and dexmedetomidine have anxiolytic properties independent of their sedative effects, but the underlying mechanisms remain unclear. We discovered a population of GABAergic neurons in the oval division of the bed nucleus of the stria terminalis that are activated by multiple anesthetics and the anxiolytic drug diazepam (ovBNSTGA). The majority of ovBNSTGA neurons express neurotensin receptor 1 (Ntsr1) and form circuits with brain regions known to regulate anxiety and stress responses. Optogenetic activation of ovBNSTGA or ovBNSTNtsr1 neurons significantly attenuated anxiety-like behaviors in both naive animals and mice with inflammatory pain, while inhibition of these cells elevated anxiety. Activation of these neurons decreased heart rate and increased heart rate variability, suggesting that they reduce anxiety by modulating autonomic responses. Our study identifies ovBNSTGA/ovBNSTNtsr1 neurons as a common neural substrate mediating the anxiolytic effect of low-dose anesthetics and a potential therapeutic target for treating anxiety-related disorders.

On cat–human interaction from the viewpoint of physics: An equation of motion

This paper provides an enjoyable example through which several concepts of classical mechanics can be understood. We introduce an equation that models the motion of a cat in the presence of a person. The cat is considered as a point particle moving in a potential induced by the person. We demonstrate that this approach to the problem reproduces characteristic behaviors of these curious animals. For instance, the fact that cats do not typically come when they are called, or that they remain longer on the lap of their favorite person; even “zoomies” are reproduced (cats randomly run back and forth across the house). We use this model problem to explore topics of current research such as stochastic equations and periodically driven systems. The pedagogical value of this work and its potential use in teaching are discussed.

Energy-Aware Feature and Classifier for Behaviour Recognition of Laying Hens in an Aviary System

Long-term monitoring of animal behaviours requires energy-aware features and classifiers to support onboard classification. However, limited studies have been conducted on the behaviour recognition of laying hens, especially in aviary systems. The objective of this study was to configure key parameters for developing onboard behaviour monitoring techniques of aviary laying hens, including proper sliding window length, energy-aware feature, and lightweight classifier. A total of 19 Jingfen No.6 laying hens were reared in an aviary system from day 30 to day 70. Six light-weight accelerometers were attached to the back of birds for behaviour monitoring with a sampling frequency of 20 Hz. Laying hen behaviours were categorized into four groups, including static behaviour (resting and standing), ingestive behaviour (feeding and drinking), walking, and jumping. Two different window lengths (0.5 s and 1 s) were tested. The SD of each axial acceleration was considered the only classification feature. The results indicated that performing denoise procedure before feature extraction can improve the classification accuracy by 10-20%. The 1-s window length yielded better accuracy than the 0.5-s window, especially for ingestive and walking behaviours. Classification models based on X-axis accelerations were better than those of Y- and Z-axis with the recognition accuracies of static, ingestive, walking, and jumping behaviours being 97.4%, 89.6%, 95.7%, and 98.5%, respectively. The study might provide insights into developing onboard behaviour recognition algorithms for laying hens.