Adaptive and maladaptive consequences of “matching habitat choice:” lessons from a rapidly-evolving butterfly metapopulation

Relationships between biased dispersal and local adaptation are currently debated. Here, I show how prior work on wild butterflies casts a novel light on this topic. “Preference” is defined as the set of likelihoods of accepting particular resources after encountering them. So defined, butterfly oviposition preferences are heritable habitat adaptations distinct from both habitat preference and biased dispersal, but influencing both processes. When a butterfly emigrates after its oviposition preference begins to reduce realized fecundity, the resulting biased dispersal is analogous to that occurring when a fish emigrates after its morphological habitat adaptations reduce its feeding rate. I illustrate preference-biased dispersal with examples from metapopulations of Melitaea cinxia and Euphydryas editha. E. editha were feeding on a well-defended host, Pedicularis, when humans created patches in which Pedicularis was killed and a less-defended host, Collinsia, was rendered phenologically available. Patch-specific natural selection favoured oviposition on Collinsia in logged (“clearing”) patches and on Pedicularis in undisturbed open forest. Quantitative variation in post-alighting oviposition preference was heritable, and evolved to be consistently different between patch types. This difference was driven more by biased dispersal than by spatial variation of natural selection. Insects developing on Collinsia in clearings retained adaptations to Pedicularis in clutch size, geotaxis and oviposition preference, forcing them to choose between emigrating in search of forest habitats with Pedicularis or staying and failing to find their preferred host. Insects that stayed suffered reduction of realized fecundity after delayed oviposition on Collinsia. Those that emigrated suffered even greater fitness penalty from consistently low offspring survival on Pedicularis. Paradoxically, most emigrants reduced both their own fitness and that of the recipient populations by dispersing from a benign natal habitat to which they were maladapted into a more demanding habitat to which they were well-adapted. “Matching habitat choice” reduced fitness when evolutionary lag rendered traditional cues unreliable in a changing environment