Mushroom body expansion and the cognitive ecology of Heliconius butterflies

Heliconius butterflies, a Neotropical genus of approximately 50 species, exhibit a marked expansion of an insect brain structure called the mushroom bodies (MBs), which are 3-4 times larger than in other Lepidoptera, including closely related Heliconiini genera. MBs are known to play a role in learning and memory, particularly in olfactory contexts, however the relative cognitive capabilities of Heliconius remain unknown. The central objective of my PhD is to investigate the behavioural consequences of, and selective pressures that drove, expansion of Heliconius MBs. I have explored these questions by collecting comparative behavioural and neuroanatomical data across Heliconius and closely related Heliconiini. To explore patterns of MB evolution in the Heliconiini, I conducted phylogenetic comparative analyses across a neuroanatomical dataset of 41 species, including 30 Heliconius and representatives of all Heliconiini genera. Phylogenetic generalised linear mixed modelling shows that within the Heliconiini, increased MB size is associated with the Heliconius genus, even when controlling for the size of the central brain, the antennal lobe and the medulla. Moreover, variable rates analyses indicate that the branch leading to Heliconius experienced a significant increase in the rate of evolution of MB size. But what drove this expansion? There are two main adaptive hypotheses to explain this MB expansion. One is that it facilitates an improved shape learning and recognition of Passiflora host plants. To test this, I conducted shape learning assays across six Heliconiini species. However, Heliconius did not, a as group, out-perform the outgroup species. In addition, I conducted geometric morphometric analyses on Passiflora leaf shape to determine the morphospace of host plants Heliconiini species exploit. There was no correlation between host plant morphospace and MB size. Together these findings suggest MB expansion in Heliconius is not associated with an improved ability for the visual identification of host plants. The second relates to Heliconius’ unique foraging strategy. Heliconius are the only Lepidoptera known to actively feed on pollen, which they collect from a limited number of relatively rare plants. In visiting these plants, Heliconius establish “traplines” – routes through the forest that they follow with a high degree of spatial and temporal fidelity, and which seemingly relies on an advanced spatial memory ability. Through a series of behavioural experiments, I show that Heliconius can learn the location of a food resource in a T-maze, in addition to outperforming non-Heliconius Heliconiini in long-term memory and non-elemental learning tasks – cognitive abilities assumed to be crucial for traplining. There was no difference, however, between Heliconius and non-Heliconius in a reversal learning task. Nonetheless, these results are consistent with the elaboration of the Heliconius MB being driven by the cognitive demands of trapline foraging for pollen. Finally, I investigate the possible neural correlates associated with long-term memory performance by comparing the mushroom bodies of Heliconius erato and Dryas iulia individuals involved in the long-term memory assay with control individuals reared in non-learning environments, in addition to a group of freshly-eclosed butterflies. Overall, the mushroom bodies of Heliconius erato exhibited significantly more age- and experience-related plasticity than Dryas iulia. Importantly, an increase in synapse count was associated directly with visual learning in Heliconius erato. At an individual level, within Heliconius erato, but not Dryas iulia, increases in synapse density and count were correlated with improved recall accuracy.