Studies of altitudinal and latitudinal gradients have identified links between the evolution of insect flight morphology, landscape structure and microclimate. Although lowland tropical rainforests offer steeper shifts in conditions between the canopy and the understorey, this vertical gradient has received far less attention. Butterflies, because of their great phenotypic plasticity, are excellent models to study selection pressures that mould flight morphology. We examined data collected over 5 years on 64 Nymphalidae butterflies in the Ecuadorian Chocó rainforest. We used phylogenetic methods to control for similarity resulting from common ancestry, and explore the relationships between species stratification and flight morphology. We hypothesized that species should show morphological adaptations related to differing micro-environments, associated with canopy and understorey. We found that butterfly species living in each stratum presented significantly different allometric slopes. Furthermore, a preference for the canopy was significantly associated with low wing area to thoracic volume ratios and high wing aspect ratios, but not with the relative distance to the wing centroid, consistent with extended use of fast flapping flight for canopy butterflies and slow gliding for the understorey. Our results suggest that microclimate differences in vertical gradients are a key factor in generating morphological diversity in flying insects.

Microhabitat choice plays a major role in shaping local patterns of biodiversity. In butterflies, stratification in flight height has an important role in maintaining community diversity. Despite its presumed importance, the role of behavioural shifts in early stages of speciation in response to differences in habitat structure is yet to be established. Here, we investigated variation in flight height behaviour in two closely related Heliconius species, H. erato cyrbia and H. himera, which produce viable hybrids but are isolated across an environmental gradient, spanning lowland wet forest to high‐altitude scrub forest. Speciation in this pair is associated with strong assortative mating, but ecological isolation and local adaptation are also considered essential for complete reproductive isolation. We quantified differences in flight height and forest structure across the environmental gradient and tested the importance of resource distribution in explaining flight behaviour. We then used common garden experiments to test whether differences in flight height reflect individual responses to resource distribution or genetically determined shifts in foraging behaviour. We found that the two species fly at different heights in the wild, and demonstrated that this can be explained by differences in the vertical distribution of plant resources. In both the wild and captivity, H. himera choose to fly lower and feed at lower positions, closely mirroring differences in resource availability in the wild. Given expectations that foraging efficiency contributes to survival and reproductive success, we suggest that foraging behaviour may reflect local adaptation to divergent forest structures. Our results highlight the potential role of habitat‐dependent divergence in behaviour during the early stages of speciation.