The application of modern statistical approaches to identify consistent individual differences in the behaviour of wild postpartum female grey seals (Halichoerus grypus)

Abstract

Consistent individual differences (CIDs) in behaviour have been shown to occur in a large number of species. However, few studies have attempted to quantify CIDs in the behaviour of wild animals in their natural environment. Yet, in order to understand the ecological and evolutionary relevance of CIDs in behaviour, it is fundamentally important that we attempt to quantify them in wild animals, in situ.

In order to address this question, a three-step analytical approach was applied to data that were collected on wild postpartum female grey seals, which were part of a ‘hands-off’ observational study. Aspects of behaviour were highly repeatable across breeding seasons. The ‘alert’ behaviour in particular, remained highly repeatable irrespective of which individuals were included in the analyses. Furthermore, these robust repeatability estimates for the ‘alert’ behaviour persisted, despite controlling for social and environmental factors that are known to influence maternal behaviour.

Subsequently, the three-step analytical approach was applied to an independent dataset collected on postpartum female grey seals that are part of a long-term longitudinal study on reproductive variation. Similar patterns were observed in the results across the three steps, and once again, the ‘alert’ behaviour was one of the more repeatable behaviours. However, these highly repeatable behaviours did not explain any of the variation in commonly used proxies for short-term fitness. As a result, these preliminary findings add to the debate on whether or not CIDs in behaviour are adaptive or are a product of phenotypic and/or genotypic constraints. Consequently, the potential influence of CIDs in behaviour on fitness trade-offs, population dynamics and conservation and management practices shall be discussed.

The application of the three-step analytical approach to the independent dataset did raise some important methodological considerations, which shall be discussed in relation to developing guidelines for applying this approach to other datasets.