Background: The thirteen species of Dryopteris in North America have long been suspected of having undergone a complicated history of reticulate evolution via allopolyploid hybridization. Various explanations for the origins of the allopolyploid taxa have been suggested, and though most lines of evidence have supported the so-called "semicristata" hypothesis, contention over the group's history has continued in several recent, conflicting studies.
Results: Sequence data from nine plastid and two nuclear markers were collected from 73 accessions representing 35 species of Dryopteris. Sequences from each of the allopolyploids are most closely related to their progenitor species as predicted by the "semicristata" hypothesis. Allotetraploid D. campyloptera appears to be derived from a hybrid between diploid D. expansa and D. intermedia; D. celsa, from diploid D. ludoviciana x D. goldiana; and D. carthusiana and D. cristata, from diploid "D. semicristata" x D. intermedia and D. ludoviciana, respectively. Allohexaploid D. clintoniana appears to be derived from D. cristata x D.goldiana. The earliest estimated dates of formation of the allopolyploids, based on divergence time analyses, were within the last 6 Ma. We found no evidence for recurrent formation of any of the allopolyploids. The sexual allopolyploid taxa are derived from crosses between parents that show intermediate levels of genetic divergence relative to all pairs of potential progenitors. In addition, the four allotetraploids are transgressive with respect to geographic range relative to one or both of their parents (their ranges extend beyond those of the parents), suggesting that ecological advantages in novel habitats or regions may promote long-term regional coexistence of the hybrid taxa with their progenitors.
Conclusions: This study provides the first thorough evaluation of the North American complex of woodferns using extensive sampling of taxa and genetic markers. Phylogenies produced from each of three datasets (one plastid and two nuclear) support the "semicristata" hypothesis, including the existence of a missing diploid progenitor, and allow us to reject all competing hypotheses. This study demonstrates the value of using multiple, biparentally inherited markers to evaluate reticulate complexes, assess the frequency of recurrent polyploidization, and determine the relative importance of introgression vs. hybridization in shaping the histories of such groups.