Combinations of correlated floral traits have arisen repeatedly across angiosperms through convergent evolution in response to pollinator selection to optimize reproduction. While some plant groups exhibit very distinct combinations of traits adapted to specific pollinators (so-called pollination syndromes), others do not. Determining how floral traits diverge across clades and whether floral traits show predictable correlations in diverse groups of flowering plants is key to determining the extent to which pollinator-mediated selection drives diversification. The North American Silene section Physolychnis is an ideal group to investigate patterns of floral evolution because it is characterized by the evolution of novel red floral color, extensive floral morphological variation, polyploidy, and exposure to a novel group of pollinators (hummingbirds). We test for correlated patterns of trait evolution that would be consistent with convergent responses to selection in the key floral traits of color and morphology. We also consider both the role of phylogenic distance and geographic overlap in explaining patterns of floral trait variation. Inconsistent with phenotypically divergent pollination syndromes, we find very little clustering of North American Silene into distinct floral morphospace. We also find little evidence that phylogenetic history or geographic overlap explain patterns of floral diversity in this group. White and pink flowering species show extensive phenotypic diversity but are entirely overlapping in morphological variation. However, red flowering species have much less phenotypic disparity and cluster tightly in floral morphospace. We find that red flowering species have evolved floral traits that align with a traditional hummingbird syndrome, but that these trait values overlap with several white and pink species as well. Our findings support the hypothesis that convergent evolution does not always proceed through comparative phenotypic divergence, but possibly through sorting of standing ancestral variation.

We took the medians of the phenotypic ranges recorded in the Flora of North America (FNA) and the Jepson eFlora, except when noted, for the following traits: stem length, leaves per node, leaf length, calyx height, calyx width, and petal limb length. We categorized the following traits: inflorescence (one, few 1-3, many >3), floral tube extension from calyx (below calyx, equal to calyx, above calyx), reproductive organ exsertion (below corolla, equal to corolla, extended above corolla). Floral color was assigned into three bins: white, pink, and red. Silene species that have white/pink floral color polymorphism were generally coded as pink to reflect the ability of the species to produce pigment in their petals. The single species that is a pale yellow, S. parishii, was coded as white to reflect the absence of anthocyanin pigments. The orange-pink S. salmonacea was coded as pink. Ploidy information was taken from a combination of sources including the FNA and the literature (Kruckeberg, 1954, Kruckeberg, 1964, Kruckeberg, 1955, Kruckeberg, 1961, Kruckeberg, 1960, Spellenberg, 1979).

Floral characters described in the FNA as ratios were calculated from the median value (e.g. “corolla 11/4 times longer than calyx height”), and were otherwise measured manually or an alternate source was sought. For species and traits that were not included or calculable in the FNA we used other flora, field guides, literature, herbarium measurements, and field measurements, see Supplementary Table 5 in the manuscript/journal article.

Any software that can open an Excel or comma-delimited file.