Abstract
Spatial mosaics occur in both evolutionary and ecological properties of species' interactions. Studies of these patterns have facilitated description and prediction of evolutionary responses of interacting species to each other and to changing environments. We propose seeking complementary understanding of community assembly and dynamics by studying ecological and mechanistic properties of mosaics. We define “species' association mosaics” as deviations from a null model in which spatial variation in the extent to which particular species interact ecologically is explained solely by variation in their densities. In extreme deviations from the null, a focal species interacts exclusively with different partners at different sites despite similar abundances of potential partners. We investigate this type of mosaic involving the butterfly Euphydryas editha and its hosts, the perennial Pedicularis semibarbata (Psem) and the ephemeral annual Collinsia torreyi (Ctor). A reciprocal transplant experiment showed that the proximate, mechanistic driver of the mosaic was variation in butterfly oviposition preference: the identity of the preferred host species depended on the site of origin of the insects, not that of the plants. In contrast, the evolutionary driver was phenological asynchrony between the insects and Ctor. Censuses showed that larvae hatching from eggs laid on Ctor would have suffered significantly greater mortality from host senescence at five sites where Ctor was avoided than at two sites where it was used. These differences among sites in phenological synchrony were caused by variation in life span of Ctor. At sites where Ctor was avoided, natural selection on host preference was stabilizing because Ctor life span was too short to accommodate the development time of most larvae. At sites where Ctor was used, selection on preference was also stabilizing because larvae lacked physiological adaptation to feed on Psem. These reciprocal forces of stabilizing selection formed a mosaic maintaining spatial variation in insect host preference that was the proximate cause of the species‐association mosaic. In the Discussion , we examine the extent to which our findings hindcast an observed anthropogenic host shift by E. editha from Psem to Ctor. This example shows that elucidation of species‐association mosaics can facilitate understanding of community evolution and dynamics.