Basal angiosperm phylogeny inferred from duplicate phytochromes A and C

We have extended our studies of angiosperm phylogeny based on a pair of duplicated phytochrome (PHY) genes, PHYA and PHYC. Phylogenetic analyses of sequences from 52 species yield unrooted gene networks in which all of the elements resolved in our previous analysis of 26 species appear. Amborella still emerges as the sister group of all other angiosperms. However, we cannot reject alternative rootings in which water lilies, either alone or in combination with Amborella, are basal. Austrobaileya + Illicium diverges next from the remaining angiosperms. Eudicots and monocots form rather well-supported clades, as do Magnoliales, Laurales, Piperales, and winteroids, but relationships among these major lineages remain uncertain, as do the positions of Chloranthaceae and Ceratophyllum. Magnoliales may be directly linked with Laurales and Piperales with winteroids, but support for these relationships is not strong. Within eudicots, a basal split between ranunculids (Ranunculales, Papaverales) and the rest of the eudicots is supported, though the position of Nelumbo is equivocal. These same relationships are obtained in combined analyses of PHYA and PHYC (species as terminals) when Ceratophyllum is excluded. However, when Ceratophyllum is included, Austrobaileya + Illicium and then Chloranthaceae diverge from the remaining angiosperms before Ceratophyllum + water lilies. Rooted species trees inferred from duplicate gene networks by minimizing gene duplications and losses are highly congruent with the gene subtrees and with the results of recent analyses of other genes, even when Ceratophyllum is included. More attention must be paid to the methods for obtaining rooted species trees from data sets that include duplicate genes, especially if we are to fully implement the search for species trees that simultaneously minimize the multiple possible causes of conflict among gene trees.