Size variation does not act as insurance in bumble bees; instead, workers add weight in an unpredictable environment

Abstract

Complex systems (e.g. eusocial insect colonies) exhibit emergent behaviours as a result of the interactions of their components. These components often vary in several traits. Such variation may improve system performance by increasing its efficiency or its robustness to environmental change. These two outcomes, efficiency and robustness, are often thought to be in opposition. Therefore, variation may be beneficial only under certain environmental conditions. Here we aim to understand why variation evolved in a particular system, bumble bee (Bombus impatiens) colonies. Workers in these colonies vary in body size, which affects the tasks they perform as well as their starvation resistance, suggesting potential impacts on efficiency and robustness, respectively. We examine how this variation affects colony performance under different environmental conditions and how colonies respond physiologically to these conditions. We maintained colonies of equal biomass but with either variable or less variable worker body sizes using targeted worker removal. We found that colonies with variable body sizes did not produce more brood (i.e. did not show evidence of increased performance) under predictable or unpredictable food environments. However, workers that developed under the unpredictable environment were smaller relative to their weight at eclosion. This effect was due in part to an increase in stored lipids, particularly in smaller workers. These physiological changes may explain why mortality rates did not differ between the predictable and unpredictable environment. Therefore, our finding that size variation did not affect colony performance suggests that size variation may be a neutral trait, present because selection is not acting against it. Our results also suggest that workers respond physiologically to differences in environmental conditions, which is important to consider when testing system robustness.