Social network analysis - centrality parameters and individual network positions of agonistic behavior in pigs over three different age levels

Genetic Associations of Novel Behaviour Traits Derived from Social Network Analysis with Growth, Feed Efficiency, and Carcass Characteristics in Pigs

Reducing harmful aggressive behaviour remains a major challenge in pig production. Social network analysis (SNA) showed the potential in providing novel behavioural traits that describe the direct and indirect role of individual pigs in pen-level aggression. Our objectives were to (1) estimate the genetic parameters of these SNA traits, and (2) quantify the genetic associations between the SNA traits and commonly used performance measures: growth, feed intake, feed efficiency, and carcass traits. The animals were video recorded for 24 h post-mixing. The observed fighting behaviour of each animal was used as input for the SNA. A Bayesian approach was performed to estimate the genetic parameters of SNA traits and their association with the performance traits. The heritability estimates for all SNA traits ranged from 0.01 to 0.35. The genetic correlations between SNA and performance traits were non-significant, except for weighted degree with hot carcass weight, and for both betweenness and closeness centrality with test daily gain, final body weight, and hot carcass weight. Our results suggest that SNA traits are amenable for selective breeding. Integrating these traits with other behaviour and performance traits may potentially help in building up future strategies for simultaneously improving welfare and performance in commercial pig farms.

Social behavior in farm animals: Applying fundamental theory to improve animal welfare

A fundamental understanding of behavior is essential to improving the welfare of billions of farm animals around the world. Despite living in an environment managed by humans, farm animals are still capable of making important behavioral decisions that influence welfare. In this review, we focus on social interactions as perhaps the most dynamic and challenging aspects of the lives of farm animals. Social stress is a leading welfare concern in livestock, and substantial variation in social behavior is seen at the individual and group level. Here, we consider how a fundamental understanding of social behavior can be used to: (i) understand agonistic and affiliative interactions in farm animals; (ii) identify how artificial environments influence social behavior and impact welfare; and (iii) provide insights into the mechanisms and development of social behavior. We conclude by highlighting opportunities to build on previous work and suggest potential fundamental hypotheses of applied relevance. Key areas for further research could include identifying the welfare benefits of socio–positive interactions, the potential impacts of disrupting important social bonds, and the role of skill in allowing farm animals to navigate competitive and positive social interactions. Such studies should provide insights to improve the welfare of farm animals, while also being applicable to other contexts, such as zoos and laboratories.

Genetic Analysis of Novel Behaviour Traits in Pigs Derived from Social Network Analysis

Social network analysis (SNA) has provided novel traits that describe the role of individual pigs in aggression. The objectives were to (1) estimate the genetic parameters for these SNA traits, (2) quantify the genetic association between SNA and skin lesion traits, and (3) investigate the possible response to selection for SNA traits on skin lesion traits. Pigs were video recorded for 24 h post-mixing. The observed fight and bullying behaviour of each animal was used as input for the SNA. Skin lesions were counted on different body parts at 24 h (SL24h) and 3 weeks (SL3wk) post-mixing. A Bayesian approach estimated the genetic parameters of SNA traits and their association with skin lesions. SNA traits were heritable (h2 = 0.09 to 0.26) and strongly genetically correlated (rg > 0.88). Positive genetic correlations were observed between all SNA traits and anterior SL24h, except for clustering coefficient. Our results suggest that selection for an index that combines the eigenvector centrality and clustering coefficient could potentially decrease SL24h and SL3wk compared to selection for each trait separately. This study provides a first step towards potential integration of SNA traits into a multi-trait selection index for improving pigs’ welfare.

Simulated poaching affects global connectivity and efficiency in social networks of African savanna elephants—An exemplar of how human disturbance impacts group-living species

Selective harvest, such as poaching, impacts group-living animals directly through mortality of individuals with desirable traits, and indirectly by altering the structure of their social networks. Understanding the relationship between disturbance-induced, structural network changes and group performance in wild animals remains an outstanding problem. To address this problem, we evaluated the immediate effect of disturbance on group sociality in African savanna elephants—an example, group-living species threatened by poaching. Drawing on static association data from ten free-ranging groups, we constructed one empirically based, population-wide network and 100 virtual networks; performed a series of experiments ‘poaching’ the oldest, socially central or random individuals; and quantified the immediate change in the theoretical indices of network connectivity and efficiency of social diffusion. Although the social networks never broke down, targeted elimination of the socially central conspecifics, regardless of age, decreased network connectivity and efficiency. These findings hint at the need to further study resilience by modeling network reorganization and interaction-mediated socioecological learning, empirical data permitting. The main contribution of our work is in quantifying connectivity together with global efficiency in multiple social networks that feature the sociodemographic diversity likely found in wild elephant populations. The basic design of our simulation makes it adaptable for hypothesis testing about the consequences of anthropogenic disturbance or lethal management on social interactions in a variety of group-living species with limited, real-world data.

An Information-Theoretic Approach to Detect the Associations of GPS-Tracked Heifers in Pasture

Sensor technologies, such as the Global Navigation Satellite System (GNSS), produce huge amounts of data by tracking animal locations with high temporal resolution. Due to this high resolution, all animals show at least some co-occurrences, and the pure presence or absence of co-occurrences is not satisfactory for social network construction. Further, tracked animal contacts contain noise due to measurement errors or random co-occurrences. To identify significant associations, null models are commonly used, but the determination of an appropriate null model for GNSS data by maintaining the autocorrelation of tracks is challenging, and the construction is time and memory consuming. Bioinformaticians encounter phylogenetic background and random noise on sequencing data. They estimate this noise directly on the data by using the average product correction procedure, a method applied to information-theoretic measures. Using Global Positioning System (GPS) data of heifers in a pasture, we performed a proof of concept that this approach can be transferred to animal science for social network construction. The approach outputs stable results for up to 30% missing data points, and the predicted associations were in line with those of the null models. The effect of different distance thresholds for contact definition was marginal, but animal activity strongly affected the network structure.

Detecting Animal Contacts-A Deep Learning-Based Pig Detection and Tracking Approach for the Quantification of Social Contacts

The identification of social interactions is of fundamental importance for animal behavioral studies, addressing numerous problems like investigating the influence of social hierarchical structures or the drivers of agonistic behavioral disorders. However, the majority of previous studies often rely on manual determination of the number and types of social encounters by direct observation which requires a large amount of personnel and economical efforts. To overcome this limitation and increase research efficiency and, thus, contribute to animal welfare in the long term, we propose in this study a framework for the automated identification of social contacts. In this framework, we apply a convolutional neural network (CNN) to detect the location and orientation of pigs within a video and track their movement trajectories over a period of time using a Kalman filter (KF) algorithm. Based on the tracking information, we automatically identify social contacts in the form of head–head and head–tail contacts. Moreover, by using the individual animal IDs, we construct a network of social contacts as the final output. We evaluated the performance of our framework based on two distinct test sets for pig detection and tracking. Consequently, we achieved a Sensitivity, Precision, and F1-score of 94.2%, 95.4%, and 95.1%, respectively, and a MOTA score of 94.4%. The findings of this study demonstrate the effectiveness of our keypoint-based tracking-by-detection strategy and can be applied to enhance animal monitoring systems.

The use of social network analysis to describe the effect of immune activation on group dynamics in pigs

The immune system can influence social motivation with potentially dire consequences for group-housed production animals, such as pigs. The aim of this study was to test the effect of a controlled immune activation in group-housed pigs, through an injection with lipopolysaccharide (LPS) and an intervention with ketoprofen on centrality parameters at the individual level. In addition, we wanted to test the effect of time relative to the injection on general network parameters in order to get a better understanding of changes in social network structures at the group level. 52 female pigs (11-12 weeks) were allocated to four treatments, comprising two injections: ketoprofen-LPS (KL), ketoprofen-saline (KS), saline-LPS (SL) and saline-saline (SS). Social behaviour with a focus on damaging behaviour was observed continuously in 10 × 15 min bouts between 0800 am and 1700 pm 1 day before (baseline) and two subsequent days after injection. Activity was scan-sampled every 5 min for 6 h after the last injection in the pen. Saliva samples were taken for cortisol analysis at baseline and at 4, 24, 48, 72 h after the injections. A controlled immune activation affected centrality parameters for ear manipulation networks at the individual level. Lipopolysaccharide-injected pigs had a lower in-degree centrality, thus, received less interactions, 2 days after the challenge. Treatment effects on tail manipulation and fighting networks were not observed at the individual level. For networks of manipulation of other body parts, in-degree centrality was positively correlated with cortisol response at 4 h and lying behaviour in the first 6 h after the challenge in LPS-injected pigs. Thus, the stronger the pigs reacted to the LPS, the more interactions they received in the subsequent days. The time in relation to injection affected general network parameters for ear manipulation and fighting networks at the group level. For ear manipulation networks, in-degree centralisation was higher on the days following injection, thus, certain individuals in the pen received more interactions than the rest of the group compared to baseline. For fighting networks, betweenness decreased on the first day after injection compared to baseline, indicating that network connectivity increased after the challenge. Networks of tail manipulation and manipulation of other body parts did not change on the days after injection at the group level. Social network analysis is a method that can potentially provide important insights into the effects of sickness on social behaviour in group-housed pigs.

Social Network Analysis in Farm Animals: Sensor-Based Approaches

Simple Summary

Social behaviour of farm animals significantly impacts management interventions in the livestock sector and, thereby, animal welfare. Evaluation and monitoring of social networks between farm animals help not only to understand the bonding and agonistic behaviours among individuals but also the interactions between the animals and the animal caretaker. The interrelationship between social and environmental conditions, and the subtle changes in the behaviours of farm animals can be understood and precisely measured only by using sensing technologies. This review aims to highlight the use of sensing technologies in the investigation of social network analysis of farm animals.

Abstract

Natural social systems within animal groups are an essential aspect of agricultural optimization and livestock management strategy. Assessing elements of animal behaviour under domesticated conditions in comparison to natural behaviours found in wild settings has the potential to address issues of animal welfare effectively, such as focusing on reproduction and production success. This review discusses and evaluates to what extent social network analysis (SNA) can be incorporated with sensor-based data collection methods, and what impact the results may have concerning welfare assessment and future farm management processes. The effectiveness and critical features of automated sensor-based technologies deployed in farms include tools for measuring animal social group interactions and the monitoring and recording of farm animal behaviour using SNA. Comparative analyses between the quality of sensor-collected data and traditional observational methods provide an enhanced understanding of the behavioural dynamics of farm animals. The effectiveness of sensor-based approaches in data collection for farm animal behaviour measurement offers unique opportunities for social network research. Sensor-enabled data in livestock SNA addresses the biological aspects of animal behaviour via remote real-time data collection, and the results both directly and indirectly influence welfare assessments, and farm management processes. Finally, we conclude with potential implications of SNA on modern animal farming for improvement of animal welfare.

Prospects for the Analysis and Reduction of Damaging Behaviour in Group-Housed Livestock, With Application to Pig Breeding

Innovations in the breeding and management of pigs are needed to improve the performance and welfare of animals raised in social groups, and in particular to minimise biting and damage to group mates. Depending on the context, social interactions between pigs can be frequent or infrequent, aggressive, or non-aggressive. Injuries or emotional distress may follow. The behaviours leading to damage to conspecifics include progeny savaging, tail, ear or vulva biting, and excessive aggression. In combination with changes in husbandry practices designed to improve living conditions, refined methods of genetic selection may be a solution reducing these behaviours. Knowledge gaps relating to lack of data and limits in statistical analyses have been identified. The originality of this paper lies in its proposal of several statistical methods for common use in analysing and predicting unwanted behaviours, and for genetic use in the breeding context. We focus on models of interaction reflecting the identity and behaviour of group mates which can be applied directly to damaging traits, social network analysis to define new and more integrative traits, and capture-recapture analysis to replace missing data by estimating the probability of behaviours. We provide the rationale for each method and suggest they should be combined for a more accurate estimation of the variation underlying damaging behaviours.

Social Network Analysis of Agonistic Behaviour and Its Association with Economically Important Traits in Pigs

Simple Summary

Aggression behaviour has several negative consequences on the performance and welfare of pigs. Here, a Social Network Analysis (SNA) approach was employed to (1) identify individual traits that describe the role of each animal in the aggression; (2) investigate the association of these traits with performance and feeding behaviour traits. The study was conducted on 326 Duroc pigs reared in 29 pens. Several individual centrality traits were identified and used to calculate the Social Rank Index. The Dominant, Subordinate, and Isolated animals represented 21.1%, 57.5% and 21.4%, respectively. No significant correlations were observed between out-degree (number of initiated agonistic behaviours) and growth traits, indicating the similarity of growth patterns for dominant and non-dominant animals. Furthermore, out-degree was correlated positively with average daily occupation time (time at the feeder/day) and average daily feeding frequency (number of visits to the feeder/day), but negatively with average daily feeding rate (gr/min). This may indicate the ability of non-dominant pigs to modify their behaviour to obtain their requirements. The Hamming distances between networks showed that there is no common behaviour pattern between pens. In conclusion, SNA showed potential for extracting behaviour traits that could be used to improve pig performance and welfare.

Abstract

Aggression behaviour has several negative consequences on the performance and welfare of pigs. Here, the Social Network Analysis (SNA) approach was employed to (1) identify individual traits that describe the role of each animal in the aggression; (2) investigate the association of these traits with performance and feeding behaviour traits. The study was conducted on 326 Duroc pigs reared in 29 pens. Several individual centrality traits were identified and used to calculate the Social Rank Index. The Dominant, Subordinate, and Isolated animals represented 21.1%, 57.5% and 21.4%, respectively. No significant correlations were observed between out-degree (number of initiated agonistic behaviours) and growth traits, indicating the similarity of growth patterns for dominant and non-dominant animals. Furthermore, out-degree was correlated positively with average daily occupation time (time at the feeder/day) and average daily feeding frequency (number of visits to the feeder/day) but negatively with average daily feeding rate (gr/min). This may indicate the ability of non-dominant pigs to modify their behaviour to obtain their requirements. The Hamming distances between networks showed that there is no common behaviour pattern between pens. In conclusion, SNA showed the potential for extracting behaviour traits that could be used to improve pig performance and welfare.

"HerdGPS-Preprocessor"-A Tool to Preprocess Herd Animal GPS Data; Applied to Evaluate Contact Structures in Loose-Housing Horses

Sensors delivering information on the position of farm animals have been widely used in precision livestock farming. Global Positioning System (GPS) sensors are already known from applications in military, private and commercial environments, and their application in animal science is increasing. However, as trade-offs between sensor cost, battery life and sensor weight have to be made, GPS based studies scheduling long data collection periods and including a high number of animals, have to deal with problems like high hardware costs and data disruption during recharging of sensors. Furthermore, human-animal interaction due to sensor changing at the end of battery life interferes with the animal behaviour under analysis. The present study thus proposes a setting to deal with these challenges and offers the software tool "HerdGPS-Preprocessor", because collecting position data from multiple animals nonstop for several weeks produces a high amount of raw data which needs to be sorted, preprocessed and provided in a suitable format per animal and day. The software tool "HerdGPS-Preprocessor" additionally outputs contact lists to enable a straight analysis of animal contacts. The software tool was exemplarily deployed for one month of daily and continuous GPS data of 40 horses in a loose-housing boarding facility in northern Germany. Contact lists were used to generate separate networks for every hour, which are then analysed with regard to the network parameter density, diameter and clique structure. Differences depending on the day and the day time could be observed. More dense networks with more and larger cliques were determined in the hours prior to the opening of additional pasture.

Winner-loser effects overrule aggressiveness during the early stages of contests between pigs

Contest behaviour, and in particular the propensity to attack an unfamiliar conspecific, is influenced by an individual's aggressiveness, as well as by experience of winning and losing (so called 'winner-loser effects'). Individuals vary in aggressiveness and susceptibility to winner-loser effects but the relationship between these drivers of contest behaviour has been poorly investigated. Here we hypothesise that the winner-loser effect on initiation of agonistic behaviour (display, non-damaging aggression, biting and mutual fighting) is influenced by aggressiveness. Pigs (n = 255) were assayed for aggressiveness (tendency to attack in resident-intruder tests) and then experienced two dyadic contests (age 10 and 13 weeks). Agonistic behaviour, up to reciprocal fighting, in contest 2 was compared between individuals of different aggressiveness in the RI test and experiences of victory or defeat in contest 1. Winner-loser effects were more influential than aggressiveness in determining initiation of agonistic behaviour. After accruing more skin lesions in contest 1, individuals were less likely to engage in escalated aggression in contest 2. The interaction between aggressiveness and winner-loser experience did not influence contest behaviour. The results suggest that aggressiveness does not compromise learning from recent contest experience and that reducing aggressiveness is unlikely to affect how animals experience winning and losing.

Agonistic Interactions in Pigs-Comparison of Dominance Indices with Parameters Derived from Social Network Analysis in Three Age Groups

Dominance indices are often calculated using the number of won and lost fights of each animal focusing on dyadic interactions. Social network analysis provides new insights into the establishment of stable group structures going beyond the dyadic approach. Thus, it was investigated whether centrality parameters describing the importance of each animal for the network are able to capture the rank order calculated by dominance indices. Therefore, two dominance indices and five centrality parameters based on two network types (initiator-receiver and winner-loser networks) were calculated regarding agonistic interactions observed in three mixing events (weaned piglets, fattening pigs, gilts). Comparing the two network types, the winner-loser networks demonstrated highly positive correlation coefficients between out-degree and outgoing closeness and the dominance indices. These results were confirmed by partial least squares structural equation modelling (PLS-SEM), i.e., about 60% of the variance of the dominance could be explained by the centrality parameters, whereby the winner-loser networks could better illustrate the dominance hierarchy with path coefficients of about 1.1 for all age groups. Thus, centrality parameters can portray the dominance hierarchy providing more detailed insights into group structure which goes beyond the dyadic approach.

A Review of the Effects of Non-Straw Enrichment on Tail Biting in Pigs

Simple Summary

Tail biting, a damaging behaviour that one pig directs at another, causes pain, wounding and health problems. It reduces both pig welfare and market value. Enrichment can reduce tail biting substantially. Many pig producers are reluctant to use straw as enrichment, but many non-straw alternatives exist. We aimed to evaluate their ability to reduce tail biting based on studies on the effects of enrichment on tail damage and manipulation of other pigs, and on the duration of interaction with enrichment. Additionally, we reviewed how pigs interact with different enrichments (e.g., by rooting or chewing it). This was done to clarify which type of enrichment could satisfy which behavioural motivation (that may lead to tail biting if not satisfied). However, very little information on separate enrichment-directed behaviours was uncovered. Several effective types of non-straw enrichment were identified, but these correspond poorly with the types of enrichment commonly applied on commercial farms. More detailed observations of how pigs interact with different enrichments, other pigs, and their environment would improve our understanding of how to combine enrichments to minimize tail biting. This is essential because although single non-straw enrichments can reduce tail biting significantly, the remaining levels of damage can still be high.

Abstract

Tail biting remains a common problem in pig production. As producers are reluctant to use straw to reduce this behaviour, we review studies on the effectiveness of other types of enrichment. Roughage, hessian sacks, compost, fresh wood, space dividers, rope, and providing new objects regularly can significantly reduce tail damage. These results should be interpreted with some caution, as often only one study per enrichment could be identified. No evidence was found that commonly applied enrichment objects (processed wood, plastic or metal) reduce tail biting significantly unless exchanged regularly, even though multiple studies per type of enrichment were identified. Many studies evaluated the duration of enrichment use, but few evaluated the manner of use. This hampers identification of combinations of enrichment that will satisfy the pig’s motivation to eat/smell, bite, root and change enrichments, which is suggested to reduce tail biting. New objects designed to satisfy specific motivations were shown to receive high levels of interaction, but their effectiveness at reducing tail damage remains unknown. More in-depth study of how pigs interact with non-straw enrichment, which motivations this satisfies and how this affects behaviour towards conspecifics, is necessary to optimize enrichment strategies. Optimization is necessary because ceasing tail docking in a way that improves pig welfare requires more effective enrichments than those described in this review, or alternatively, better control over other factors influencing tail biting.

Social network analysis in pigs: impacts of significant dyads on general network and centrality parameters

In general, one animal is considered dominant over another animal if it has won more fights than its opponent. Whether this difference in won and lost fights is significant is neglected in most studies. Thus, the present study evaluates the impact of two different calculation methods for dyadic interactions with a significant asymmetric outcome on the results of social network analysis regarding agonistic interactions of pigs in three different mixing events (weaned piglets, fattening pigs and gilts). Directly after mixing, all animals were video recorded for 17 (fattening pigs, gilts) and 28 h (weaned piglets), documenting agonistic interactions. Two calculation methods for significant dyads, that is, dyadic interactions with a clear dominant subordinate relationship in which one animal has won significantly more fights than its encounter, were proposed: pen individual limits were calculated by a sign test considering the differences of won and lost fights of all dyadic interactions in each pen; dyad individual limits were determined by a one-sided sign test for each individual dyad. For all data sets (ALL, including all dyadic interactions; PEN or DYAD, including only significant dyads according to pen or dyad individual limits), networks were built based on the information of initiator and receiver with the pigs as nodes and the edges between them illustrating attacks. General network parameters describing the whole network structure and centrality parameters describing the position of each animal in the network were calculated. Both pen and dyad individual limits revealed only a small percentage of significant dyads for weaned piglets (12.4% or 8.8%), fattening pigs (4.2% or 0.6%) and gilts (3.6% or 0.4%). The comparison between the data sets revealed only high Spearman’s rank correlation coefficients (r_S) for the density, that is, percentage of possible edges that were actually present in the network, whereas the centrality parameters showed only moderate r_S values (0.37 to 0.75). Thus, the rank order of the animals changed due to the exclusion of insignificant dyads, which shows that the results obtained from social network analysis are clearly influenced if insignificant dyads are excluded from the analyses. Due to the fact that the pen individual limits consider the overall level of agonistic interactions within each pen, this calculation method should be preferred over the dyad individual limits. Otherwise, too many animals in the group became isolated nodes with zero centrality for which no statement about their position within the network can be made.

Impact of Significant Dyads on Dominance Indices in Pigs

Dominance indices are calculated by considering the differences between the number of won and lost fights. Whether these differences show a significant asymmetric outcome or not is neglected. Thus, two calculation methods for the limits of significant dyads are proposed using a sign test based on the differences in won and lost fights, considering all dyadic interactions in the pen (PEN: pen individual limits), and a sign test focusing on each individual dyad (DYAD: dyad individual limits). These were compared to the data set containing all dyadic interactions (ALL). Agonistic interactions in three mixing events were video recorded for two and a half days (weaned piglets) or one and a half days (fattening pigs, gilts). Dominance indices (DI) were calculated for all data sets. Pen/dyad individual limits revealed a small number of significant dyads (weaned piglets: 12.4%/8.8%; fattening pigs: 4.2%/0.6%; gilts: 3.6%/0.4%). Pen individual limits should be selected as they allow adaption of the limits according to the fighting frequency. Spearman rank correlation coefficients of the dominance indices between the data sets were always above 0.7, implying that the rank order remained relatively stable. Information about the impact of significant dyads on sociometric measures is important to prevent misinterpretations about the social structure in animal groups and should be considered in future studies.

A systematic survey of centrality measures for protein-protein interaction networks

Background

Numerous centrality measures have been introduced to identify “central” nodes in large networks. The availability of a wide range of measures for ranking influential nodes leaves the user to decide which measure may best suit the analysis of a given network. The choice of a suitable measure is furthermore complicated by the impact of the network topology on ranking influential nodes by centrality measures. To approach this problem systematically, we examined the centrality profile of nodes of yeast protein-protein interaction networks (PPINs) in order to detect which centrality measure is succeeding in predicting influential proteins. We studied how different topological network features are reflected in a large set of commonly used centrality measures.

Results

We used yeast PPINs to compare 27 common of centrality measures. The measures characterize and assort influential nodes of the networks. We applied principal component analysis (PCA) and hierarchical clustering and found that the most informative measures depend on the network’s topology. Interestingly, some measures had a high level of contribution in comparison to others in all PPINs, namely Latora closeness, Decay, Lin, Freeman closeness, Diffusion, Residual closeness and Average distance centralities.

Conclusions

The choice of a suitable set of centrality measures is crucial for inferring important functional properties of a network. We concluded that undertaking data reduction using unsupervised machine learning methods helps to choose appropriate variables (centrality measures). Hence, we proposed identifying the contribution proportions of the centrality measures with PCA as a prerequisite step of network analysis before inferring functional consequences, e.g., essentiality of a node.

Electronic supplementary material

The online version of this article (10.1186/s12918-018-0598-2) contains supplementary material, which is available to authorized users.