Inaccurate and unverified information in decision making: a model for the nest site selection process of 'Apis florea'

Abstract

We present an agent-based model for the nest site selection process of the open-nesting red dwarf honeybee, 'Apis florea'. Our main aim was to determine how nest site requirements affect the bees' decision-making process. We either calculated our model parameters from experimental data or chose them so that our model would generate similar numbers of dancing bees and dance followers to those observed in real swarms with access to an abundance of suitable nest sites in all directions. We found that 'A. florea' is less capable of making a collective decision on a new nest site when the area occupied by suitable sites is small compared to when suitable sites are abundant. Increasing the use of information regarding the location of potential nest sites or the accuracy of the information available enhanced the decision-making ability of 'A. florea' when nest sites were scarce. We also found that swarm guidance might be hindered when suitable nest site areas are wide apart. We therefore examined two possible mechanisms for increasing directional agreement among dancers: mimicry of unverified dance information and self-regulation by inhibiting or changing dance behaviour based on observations of other dances. We show that, even at low levels, dance mimicry greatly enhances the ability of an 'A. florea' swarm to make a decision and reduces the time to make a decision. However, in the presence of mimicry errors propagate through the swarm. Self-regulation had little or no effect, probably because of the overall low levels of dance activity present on the swarm at any given time. Our model results suggest that 'A. florea's' decision-making process allows swarms to locate a new nest site provided nest sites are abundant, even when they are of similar quality.