Abstract
Animal behaviour often is characterised by standardised assays. In social insects such as ants, behaviour assays are for example used to characterise aggressive and peaceful behaviour. Such assays differ in the number of individuals, the duration and place of assays, and the scoring scales. Also the behaviour indices used to summarise the results differ. Here, we compared five behaviour indices (Aggression Index, Mean Maximum Aggression Index; and the newly introduced Mean Maximum Peace Index, Mean Behaviour Index aggressive, and Mean Behaviour Index peaceful) using a scoring scale that comprises peaceful and aggressive behaviour. The indices were applied on eight simulations and three observed data sets. The five indices were correlated but frequently differed in their means. Multiple indices were needed to capture the complete behaviour range. Furthermore, subtle differences in workers' behaviour, that is, differences that go beyond the presence/absence of aggression, were only identified when considering multiple indices. We infer that the indices applied are differently suited for different analyses. Fine-scale analyses of behavioural variation profit from using more than one index. The particular choice of index or indices likely influences the interpretation of behaviour and should be carefully done in the light of study species and research question.