Abstract

This investigation on stereotypies of domestic sows has two aims: 1) to investigate the behavioural profiles shown in the different parities and stages of the reproductive cycle, and 2) to characterise stereotypies by descriptive qualitative and quantitative parameters. As the commonly used functional characteristics of stereotypies were felt to be largely hypothetical at this stage, preventing a deeper understanding of this abnormal behaviour, two purely descriptive approaches were adopted: the first one separated such behaviour from other activities by intuitively essential features, that were condensed into an operational definition. Observations were made on the intensively kept sows of the Edinburgh School of Agriculture pig unit kept in the dry sow house and the lactating sow stalls. From scanning surveys and continuous focal animal samples, behavioural profiles were obtained for the sows of different ages and physiological states. Parity 1 was characterised by frequent and long lasting drowsy stances which probably were a "cut-off" reaction to the unfamiliar Dry-Sow-House environment. In parity 2 and 3 the sows showed increased investigative and manipulative behaviour, but also rapidly emerging stereotyped activities. These increased in kind, frequency and duration over the parities, but were confined to the pregnant animals in the dry sow stalls and virtually absent from the lactating sows which could interact with their piglets. The second approach was to measure the general repetitivity in the behavioural sequences by quantifying their informational content with the redundancy measure C. Both appraoches largely agreed when the intuitively defined stereotyped or non-stereotyped sections of behaviour were assessed for redundancy, the former being significantly more redundant in spite of overlapping C-values. The redundancy of all behaviour, summing variable, positional and stereotyped activities, gradually rose with parity through all cycle stages. In part this was due to the increasing proportion of the intuitively stereotyped sequences, and in part to the increasing redundancy of the remaining variable activity. The redundancy in the stereotyped sequences however remained fairly constant once they were established, and were in all parities well above the levels of variable behaviour. In order to appropriately compare stereotyped and variable activity, redundancies were not only calculated relative to the entire range of elements observed in the stalled dry sows, but also relative to the restricted range of the active elements of all or of each individual sow. All ANOVAs showed significant increases of redundancy over parities, but no effects of pregnancy or interaction to parity. In the discussion the two approaches are compared; from the process of acquiring the sterotypies we then derived the hypothesis, that a likely function could be to reach homeostasis in arousal by warding-off external stimuli and self-generating known sensory input. So far there is nevertheless firm evidence, that stereotypies are signs of major behavioural problems and of concurrently impaired welfare, especially at times of acquisition.