26The biotic and the abiotic environment play a major role in shaping plant 27 phenotypes and their geographic distributions. However, little is known about the 28 extent to which plant phenotypes match local patterns of herbivory across fine-29 grained habitat mosaics, despite the strong effect of herbivory on plant fitness. 30 Through a reciprocal transplant-common garden experiment with clonally 31 propagated rhizomes, we tested for local phenotypic differentiation in bittercress 32 (Brassicaceae: Cardamine cordifolia) plants collected across an ecotonal habitat 33 mosaic. We found that bittercress in sunny meadows (high herbivory) and shaded 34 understories (low herbivory) have diverged in heritable growth and herbivore 35 resistance phenotypes. The expression of these differences was habitat dependent, 36 mirroring patterns of adaptive divergence in phenotypic plasticity between plant 37 populations in meadow and understory habitats at broader geographic scales, and 38 showed no evidence for a constraint imposed by growth-defense tradeoffs. Most 39 notably, plants derived from shade habitats exhibited a weaker shade-induced 40 elongation response (i.e., shade avoidance syndrome, SAS) and reduced resistance 41 to herbivory, relative to plants derived from sun habitats, when both were grown in 42 shade common gardens. Greenhouse experiments revealed that divergent SAS 43 phenotypes in shade conditions were expressed in offspring grown from seed as well.
44Finally, we observed partially non-overlapping flowering phenology between 45 habitat-types in the field, which may be at least one factor that helps to reinforce 46 habitat-specific phenotypic divergence. Altogether, our study illuminates how a 47 48 grained habitat mosaic. 49 50 51 55 2005; Richardson et al. 2014). However, our knowledge regarding the conditions 56 under which these patterns arise and persist has been disproportionately shaped by 57 studies at course-grained rather than fine-grained spatial scales (Richardson et al.58 2014). This bias has likely been shaped by the expectation that gene flow among 59 interspersed habitat patches is generally a strong homogenizing force, preventing 60 the establishment of habitat-associated phenotypic and genotypic variation at fine-61 grained spatial scales (Haldane 1930; Lenormand 2002). Although there is growing 62 evidence that heritable phenotypes track habitat mosaics at fine-grained spatial 63 3 scales [i.e. microgeographic phenotypic divergence; (Richardson et al. 2014)] we 64 have a limited understanding of the molecular, ecological, and evolutionary 65 processes that facilitate and maintain this variation in nature. 66 67 Defoliation by insect herbivores exerts strong selection on plant phenotypes (Louda 68 1984; Prasad et al. 2012; Agrawal et al. 2012). In mustards (Brassicaceae), 69 polymorphisms in genes that modify defensive chemicals underlie adaptation to 70 local herbivore communities (Prasad et al. 2012; Zust et al. 2012), and the 71 magnitude of geographic divergence at such loci is extr...