Contribution of maternal effects to dietary selection in Mediterranean fruit flies.

Individual responses to dietary variation represent a fundamental component of fitness, and nutritional adaptation can occur over just a few generations. Maternal effects can show marked proximate responses to nutrition, but whether they contribute to longer term dietary adaptation is unclear. Here, we tested the hypotheses that maternal effects: (i) contribute to dietary adaptation, (ii) diminish when dietary conditions are constant between generations, (iii) are trait-specific and (iv) interact with high- and low-quality food. We used experimental evolution regimes in the medfly (Ceratitis capitata) to test these predictions by subjecting an outbred laboratory-adapted population to replicated experimental evolution on either constant high calorie sugar ('A') or low-calorie starch ('S') larval diets, with a standard adult diet across both regimes. We measured the contribution of maternal effects by comparing developmental and adult phenotypes of individuals reared on their own diet with those swapped onto the opposite diet for either one or two generations (high and low maternal effect conditions, respectively), both at the start and after 30 generations of selection. Initially, there were strong maternal effects on female body mass and male mating success but not larval survival. Interestingly, the initial maternal effects observed in female body mass and male mating success showed sex-specific interactions when individuals from high calorie regimes were tested on low calorie diets. However, as populations responded to selection, the effects of maternal provisioning on all traits diminished. The results broadly supported the predictions. They show how the contribution of maternal effects to dietary responses evolves in a context-dependent manner, with significant variation across different fitness-related traits. We conclude that maternal effects can evolve during nutritional adaptation and hence may be an important life history trait to measure, rather than to routinely minimize.