Diapause, the temporary cessation of development at an early life-history stage, is widespread among animals and plants. The range of taxa exhibiting various forms of diapause indicates its enormous ecological significance and highlights its value as a model for examining life-history trait evolution. However, despite the impact of diapause on species ecology, there is little understanding of its adaptive value in many groups. Furthermore, the relative roles of phylogeny and ecology in determining the contemporary expression of the trait remain unresolved. Delayed implantation (DI) is a type of diapause found in several orders of mammals. It is particularly prevalent in the Mustelidae, with mustelids making up more than half of all mammals known to exhibit DI. This taxon is thus ideal for examining life-history predictors of DI and investigating the mode of evolution. Both maximum likelihood and maximum parsimony methods of ancestral state reconstruction indicated DI to be plesiomorphic in the mustelids, although multiple state changes are required to explain its contemporary distribution. After controlling for phylogeny, species with and without DI could be discriminated using just three variables: longevity, maximum latitude of the geographical distribution, and a term describing maternal investment. Our analyses supported the hypothesis that DI is more prevalent in seasonal climates. We also showed that longer-lived species are more likely to exhibit DI, suggesting a time cost to the trait. We found no correlate for the highly variable duration of DI, which remains unexplained. Although ecological factors can predict the distribution of DI in modern mustelids, phylogenetic constraint is likely to play an important role.

Adaptation to one novel host plant may simultaneously improve an insect's performance on other unfamiliar hosts, as a kind of cross-adaptation. In selection experiments using an Asian population of the seed beetle Callosobruchus maculatus (Fabricius) (Coleoptera: Chrysomelidae: Bruchinae), larval survival in lentil seeds increased from about 1 to >90% in fewer than 30 generations, and acceptance of lentil by egg-laying females increased two-to three-fold. We examined whether rapid adaptation to lentil altered beetle performance on other grain legumes, including hosts more closely related to lentil than to the ancestral host, mung bean. Three replicate, lentil-adapted lines were compared to the mung-bean (control) line with respect to both host acceptance (oviposition under no-choice conditions) and larval performance (survival to adult emergence, development time, and adult mass at emergence). In most experiments, females from the lentil lines laid more eggs on unfamiliar hosts than did females from the mung-bean line. Greater oviposition on novel hosts could not be explained by variation in potential fecundity, and did not appear to depend on a host's relatedness to lentil. Although survival in lentil remained extremely divergent between the mung-bean and lentil lines (0 vs. >90%), the lines did not differ in larval performance on two other novel hosts (pea and fava bean) that are much more closely related to lentil than to mung bean. Because larval performance is most likely the limiting factor in the potential colonization of a novel host by C. maculatus, our experiments did not provide strong evidence for cross-adaptation. The results suggest that adaptation to even a highly marginal host need not lead to a general expansion of an insect's host range.