Radial arm maze as a new paradigm to study collective behaviours in fish

Collective decision-making is based on both environmental information perceived by individuals and social interactions with other group members. Determining and analyzing separately both interactions is a real challenge. If the environmental influences on group behaviours can be determined, new possibilities to collect information about processes inside the group become possible. To improve our knowledge of these processes, an experiment where collective decision-making can be measured easily and without any ambiguity is needed. For this perspective, a new paradigm in the study of collective behaviour is introduced here. The radial arm maze is a classical method used to study individual cognitive abilities. Its advantages are firstly to allow control of environmental information; secondly, to realize multi-way tests, and thirdly, to give the opportunity to collect categorical responses like presence/absence. We apply this paradigm for the first time to a whole animal group. We have also developed an image analysis system able to automatically count the number of individuals in every defined zone. Due to this counting, the degree of cohesion, the group stability, the activity and zone preferences can be described as function of factors such as the group size, the defined zones, or the experimental time. The degree of cohesion can be measured by a new index taking into account the number of sub-groups and the size of each ones. Group activity can be measured by the movement of the majority group between arms. This activity allows determining exploratory processes but also whether zone preferences or homing phenomena appear in the absence of any stimulus. To illustrate, our first results from the exploratory behaviour study of shoals of golden shiners (Notemigonus crysoleucas) are introduced. Using this new paradigm, it is now possible to quantify rapidly in a standardised way the collective responses of fish shoals according to the absence or presence of environmental stimuli, and to create experiments where environmental information is controlled