Abstract
Animal personality is defined as individual variation in behaviour which is consistent over time and across contexts; the five major components of personality are activity, boldness, exploration, sociability, and aggression. Many taxa show sex differences in behaviours, including in personality. Most studies do not assess repeatability of behaviour, and if they do, they generally assess repeatability over only relatively short time periods (e.g., a week or a month). Understanding whether personality traits are repeatable over longer periods is important for the assessment of individual differences in behaviour; if a behaviour is not repeatable over time, then it is uninformative about consistent behavioural syndromes within individuals. A core notion of personality research is that behaviours are consistent in different contexts, but there is also evidence of behaviour changing within individuals with a change in certain contexts. Thus, understanding which contexts cause a change in behaviour and which do not may be important for future research that assumes consistency of behaviour. In this thesis, I use zebrafish (Danio rerio) as a model to assess behavioural consistency over time and context.
In Chapter 2, I investigate the repeatability of activity, exploration, boldness, anxiety, and sociability to obtain longer-term (6 month) estimates of repeatability. I hypothesised: that these behaviours would be repeatable; that there would be sex differences both in repeatability estimates and in actual values; and, finally, that behavioural syndromes would be expressed. Specifically, this required that the expression of one personality trait would be correlated with another. To answer these questions, individuals were filmed in a test sequence, comprising a novel arena assay (to test activity, anxiety and exploration) and a novel object assay within the same tank (to test boldness). Single-sex groups were then filmed in a different arena to test sociability. Behaviours were quantified using the behavioural analysis programs EthoVision XT and idTracker. Overall, activity and exploration in individuals, and sociability at the tank level, were fairly repeatable over six months. Boldness was also repeatable, but to a lesser degree. A commonly-used measure of anxiety was more repeatable in the novel arena assay than the novel object test. There were clear sex differences in aspects of activity and sociability, in that males were more active and shoaled less tightly than females. However, while males were more repeatable in exploratory behaviour, there were minimal sex differences in the mean expression of exploration and boldness. There was evidence for an activity-boldness-exploration behavioural syndrome, but conclusions drawn from this part of the analysis are tentative because of the non- independence of the activity and exploration data.
After demonstrating that anxiety over the first ten minutes in a novel tank is a repeatable behaviour in zebrafish, in Chapter 3 I assessed how a behavioural challenge, like social isolation, affects anxious behaviour. In social animals, isolation from conspecifics can negatively affect both behaviour and physiology. It has been previously demonstrated that isolating zebrafish from conspecifics can induce them to become more anxious and aggressive, but I aimed to investigate whether individuals that were more or less anxious than the norm would react differently, and whether their anxiety levels would remain similar to their initial behaviour. I predicted that, after a three-week period of isolation, initially non-anxious zebrafish would become more anxious, and anxious fish would remain so. Using only the extreme anxious and non-anxious fish, I kept each individual either isolated or in a group of five for a three-week period, and then phenotyped these same fish again to get a post-treatment estimate of anxiety. I found that all four treatment groups (anxious or non- anxious, and isolated or grouped) showed a mean change in anxiety, but while the anxious isolated and group-housed fish remained on the anxious side of the spectrum and the non- anxious group-housed fish remained on the non-anxious side, the non-anxious isolated fish showed double the change of any other group, and were (on average) classed as mostly anxious after the treatment. My sample sizes were small, and so my model did not detect an overall significant change, but the effect sizes indicated a true effect of treatment despite the low power of my study. This supports part of my hypothesis that non-anxious fish would become more anxious after a period of isolation.
The major findings from this study were that activity- and exploration-related (and, to a lesser degree, boldness-related) behaviours were repeatable over a six-month period, and that social isolation is enough of a behavioural challenge to alter previous anxiety levels, especially in initially non-anxious fish. I also found evidence of both sex differences and behavioural syndromes over time. The research done in this thesis contributes to the established literature by generating long-term behavioural repeatability estimates for a model species, as well as an experimental assessment of how social isolation can alter anxiety levels. Future research should evaluate the applicability of behavioural tests to the behaviours they intend to study.