Phenotypic plasticity, the ability of a genotype to express different phenotypes in different environments, allows organisms to cope with variation in resources and invade novel environments. Biologists have long been fascinated with the costs and tradeoffs that generate and maintain variation in plasticity, such as possible increases in brain size and delays in reproduction associated with the evolution of learning. However, the costs of plasticity vary: many studies have failed to find costs of plasticity, the degree of costs often vary with the system or environments considered, and many costs of plasticity are variable even within the lifetime of an individual. This research adopts a developmental perspective to predict the degree and incidence of costs of plasticity, using host learning in butterflies as a case study. Learning, a mechanism of plasticity that develops through a trial-and-error sampling process, should result in developmental costs and allocation of energy towards development (at the expense of reproduction). Furthermore, costs of learning should be less pronounced in environments for which organisms have innate biases and for learned traits underlain by short-term memory, relative to long-term memory (which requires more developmental re-structuring). This research found support for all three predictions across three levels of costs: behavioral costs, tissue costs, and fecundity trade-offs. Butterflies exhibited genetic variation in their ability to learn to recognize different colored hosts. Genotypes with higher proxies for long-term memory emerged with relatively larger neural investment and smaller reproductive investment. In contrast to these costs of long-term learning, proxies of short-term learning were only correlated with increased exploration of a range of possible resources (types of non-hosts) early in the host-learning process. Family-level costs of plasticity emerged from the ability to learn to locate a red host, for which butterflies do not have an innate bias. Costs of learning were also induced by learning itself: following exposure to novel (red) host environments, individual butterflies, regardless of genetic background, increased exploratory behavior, increased neural investment, and re-allocated energy away from reproduction towards other functions (e.g., flight). Considering developmental mechanisms helps to predict how costs will influence the evolution of learning and plasticity
