Affective and cognitive components of agonistic behaviour in pigs: implications for animal welfare and social behaviour

Abstract

Individuals differ in their propensity towards aggressive behaviour and the causes of this variation are poorly understood. In commercial pig farming systems, pigs negotiate a complex and costly series of agonistic interactions, which are likely to be cognitively demanding, to stimulate strong emotions, and to shape future social behaviour. Understanding how cognitive ability and affective response to aggression determines aggressive behaviour enables us to quantify the emotional impact of aggression and identify avenues for improving pig welfare. I examine the proximate causes of aggression in dyadic contests and groups of unfamiliar conspecifics of the same age and similar size; a social environment commonly encountered by commercially farmed pigs. Using social network analysis allowed quantification of the importance of group-level and indirect connections to the aggression performed and received. Chapter 2 reviews how social network analysis has been applied to farm animal sociality.

While key centrality metrics have been characterised, chiefly in cattle and pigs, there is a need for replicated networks, measurement of welfare outcomes and dynamic network analysis to identify causal links between social network traits and welfare. This will allow us to assess the effect of an animal’s social environment as a whole on welfare outcomes and, ensure that animals that are managed in groups are benefitting from appropriate companionship. Chapter 3 explores how the interplay between intrinsic and extrinsic influences on aggression affect how aggression is expressed. Chapter 3 assessed the relative importance of aggressiveness as a personality trait, the experience of winning and losing, and their interaction, on the subsequent initiation of agonistic behaviour. Tendency to attack in two resident intruder tests was assayed at 9 weeks of age as a measure of aggressiveness (‘RI aggressiveness’). The residents then experienced two dyadic contests, at age 10 and 13 weeks.

Agonistic behaviour in the second contest, up to reciprocal fighting, was compared between individuals of different RI aggressiveness and experiences of victory or defeat in the first contest. Whilst greater RI aggressiveness predicted faster initiation of agonistic behaviour in the pre-escalation period, winner-loser effects were more influential than RI aggressiveness in determining initiation of agonistic behaviour. This led to the conclusion that aggressive personality did not prevent pigs from learning from winning and defeat. Therefore, the persistent, escalated aggression that occurs after pigs are mixed is not explained by the inflexible application of aggression by highly aggressive individuals. Further research into how animals vary in their ability to learn quickly and efficiently from their experience of aggression is explored using cognitive tests in chapters 5 and 6.

Chapter 4 explores the emotional impact of dyadic aggression on behaviour in a subsequent contest. A combination of qualitative and quantitative methods was used to assess the emotional expression of pigs in the pre-escalation period of the dyadic contests analysed in chapter 1. Qualitative behavioural assessment (QBA) identified three principal components: PC1: agitated/tense to relaxed/content, PC2: fearful/aimless to confident/sociable and PC3: listless/ indifferent. Less aggressive pigs and males showed a more positive emotionality (PC2) whilst PC1 and PC3 were unaffected by first contest outcome and aggressiveness. More aggressive pigs in the resident-intruder test were more likely to hold their ears back during the early contest period. Approaching the opponent was influenced by winning and losing a prior contest in the low aggressive pigs only. Less aggressive pigs may be more susceptible to the emotional impact of victory and defeat but overall, and more aggressive pigs express more negative emotionality at the start of agonistic encounters. The link between negative emotions and the initiation of aggression is not well understood in pigs, but this study concludes that high aggressiveness is not explained by fearlessness.

The second part of the thesis investigates whether aggression is associated with performance in a two-part cognitive test. Pigs learnt to discriminate between two locations, one rewarded with food and the other punished with a fan turning on suddenly. Afterwards, the rewarded and punished locations were reversed, to challenge pigs’ cognitive flexibility and their ability to inhibit their previously learnt association. Chapter 5 tested the effect of cognitive ability on aggression at the pen level, during the peak of group aggression, for 5 hours immediately post-regrouping. Groups were formed by taking the best performing two of four littermates in the cognitive tests into a “high” test performance pen, and the worst two to a “low” test performance pen, forming pens of median 14 pigs from 7 litters. High and low test performance pens did not differ in density of network connections or how aggression was centralised within the pen (degree, betweenness and eigenvector centralisation) or the size of the largest clique.

Aggression was less centralised in pens containing pigs with a higher pass rate in the reversal learning test, and this applied to “all aggression”, unilateral aggression and mutual fighting networks. The final chapter assesses whether cognitive ability affects pigs’ roles in aggression networks. Spatial discrimination ability did not predict individual centrality but passing the reversal learning test significantly increased weighted in- degree in unilateral aggression networks.

This means that pigs that performed better at reversal learning received more unilateral aggression that pigs that failed the test. “All-aggression” and mutual fighting networks were not predicted by cognitive tests. Enhanced cognitive ability did not appear to reduce total aggression or skin lesions. This may reflect limitations of testing cognition in a non-social context or reflect the constraints of mixing pigs in a confined environment, which limits how effectively pigs can avoid aggression, regardless of their cognitive ability.

In summary, pigs engage in costly aggressive interactions despite experiencing negative emotions associated with aggression. Pigs adapt their aggressive behaviour according to experience, showing that even highly aggressive pigs can learn from victory and defeat. Performance in reversal learning tests appears to be more strongly related to the structure of aggression in newly formed groups of pigs, than spatial discrimination. A focus on developing how the social domain of cognition develops in pigs may be required to establish what mechanisms link cognitive ability with aggression.