Summary Altered precipitation patterns associated with anthropogenic climate change are expected to have many effects on plants and insect pollinators, but it is unknown if effects on pollination are mediated by changes in water availability. We tested the hypothesis that impacts of climate on plant–pollinator interactions operate through changes in water availability, and specifically that such effects occur through alteration of floral attractants. We manipulated water availability in two naturally occurring Mertensia ciliata (Boraginaceae) populations using water addition, water reduction and control plots and measured effects on vegetative and floral traits, pollinator visitation and seed set. While most floral trait values, including corolla size and nectar, increased linearly with increasing water availability, in this bumblebee‐pollinated species, pollinator visitation peaked at intermediate water levels. Visitation also peaked at an intermediate corolla length, while its relationship to corolla width varied across sites. Seed set, however, increased linearly with water. These results demonstrate the potential for changes in water availability to impact plant–pollinator interactions through pollinator responses to differences in floral attractants, and that the effects of water on pollinator visitation can be nonlinear. Plant responses to changes in resource availability may be an important mechanism by which climate change will affect species interactions.

No abstract

Summary Land managers choose seed from a variety of provenances for restoration projects. By selecting seed of the local ecotype, managers can increase establishment in the short term and prevent the disruption of local adaptations and genetic swamping in the long term. However, local seed may be disadvantageous if populations are inbred or maladapted to managed restoration environments. Seed selection may be further confounded by propagation methods. Three dominant C4 grasses, Andropogon gerardii, Bouteloua curtipendula and Sorghastrum nutans, from three types of seed provenances (remnant, restoration and nursery) were planted as seeds and plugs into experimental plots at three established tallgrass prairie restorations in western Minnesota, USA. Using a common garden design, we tested whether (i) provenance and (ii) site of planting influence germination and first‐season survival and growth both (iii) for seeds directly planted in the field and for transplants (plugs). Seed provenance impacted germination and seedling survival in all cases, except S. nutans seeded directly in the field. Andropogon gerardii and B. curtipendula nursery seedlings were consistently taller than those of the other provenance types. When directly seeded, germination, survival and vigour differed among restoration sites; however, the results were species specific. Sorghastrum nutans germination varied among sites depending on provenance, indicating that this species may be particularly sensitive to environmental conditions. Germination was 3–12 times greater for plugs than for seeds directly planted in the field in summer, but mortality after planting in the field was low for both groups. Synthesis and applications. Provenance and restoration site had varying effects among species, indicating that the dominant C4 grasses used in this study ought not to be considered ecological equivalents. While we found little evidence of local adaptation, use of local remnant seeds diminishes the risk of spreading maladapted genotypes. Germination limited establishment when sowing seeds directly in summer. Supplemental use of plugs may increase species diversity in restorations.