Corolla length is a labile flower feature and has strong implications for pollinator success. However, the phenotypic and genetic bases of corolla elongation are not well known, largely due to a lack of good candidate genes for potential genetic exploration and functional work. We investigate both the cellular phenotypic differences in corolla length, as well as the genetic control of this trait, in Saltugilia (Polemoniaceae). Taxa in this clade exhibit a large range of flower sizes and differ dramatically in pollinator guilds. Flowers of each species were collected from multiple individuals during four stages of flower development to ascertain if cell number or cell size is more important in determining flower size. In Saltugilia, increased flower size during development appears to be driven more by cell size than cell number. Differences in flower size between species are governed by both cell size and cell number, with the large-flowered S. splendens subsp. grantii having nearly twice as many cells as the small-flowered species. Fully mature flowers of all taxa contain jigsaw cells similar to cells seen in sepals and leaves; however, these cells are not typically found in the developing flowers of most species. The proportion of this cell type in mature flowers appears to have substantial implications, comprising 17–68% of the overall flower size. To identify candidate genes responsible for differences in cell area and cell type, transcriptomes were generated for two individuals of the species with the smallest (S. australis) and largest (S. splendens subsp. grantii) flowers across the same four developmental stages visualized with confocal microscopy. Analyses identified genes associated with cell wall formation that are up-regulated in the mature flower stage compared to mid-stage flowers (75% of mature size). This developmental change is associated with the origin of jigsaw cells in the corolla tube of mature flowers. Further comparisons between mature flowers in the two species revealed 354 transcripts that are up-regulated in the large-flowered S. splendens subsp. grantii compared to the small-flowered S. australis. These results are likely broadly applicable to Polemoniaceae, a clade of nearly 400 species, with extensive variation in floral form and shape.

Premise of the study:Visualizing flower epidermal cells is often desirable for investigating the interaction between flowers and their pollinators, in addition to the broader range of ecological interactions in which flowers are involved. We developed a protocol for visualizing petal epidermal cells without the limitations of the commonly used method of scanning electron microscopy (SEM).Methods:Flower material was collected and fixed in glutaraldehyde, followed by dehydration in an ethanol series. Flowers were dissected to collect petals, and subjected to a Histo-Clear series to remove the cuticle. Material was then stained with aniline blue, mounted on microscope slides, and imaged using a compound fluorescence microscope to obtain optical sections that were reconstructed into a 3D image.Results:This optical sectioning method yielded high-quality images of the petal epidermal cells with virtually no damage to cells. Flowers were processed in larger batches than are possible using common SEM methods. Also, flower size was not a limiting factor as often observed in SEM studies. Flowers up to 5 cm in length were processed and mounted for visualization.Conclusions:This method requires no special equipment for sample preparation prior to imaging and should be seen as an alternative method to SEM.