A duplicate gene rooting of seed plants and the phylogenetic position of flowering plants

Abstract

Flowering plants represent the most significant branch in the tree of land plants, with respect to the number of extant species, their impact on the shaping of modern ecosystems and their economic importance. However, unlike so many persistent phylogenetic problems that have yielded to insights from DNA sequence data, the mystery surrounding the origin of angiosperms has deepened with the advent and advance of molecular systematics. Strong statistical support for competing hypotheses and recent novel trees from molecular data suggest that the accuracy of current molecular trees requires further testing. Analyses of phytochrome amino acids using a duplicate gene-rooting approach yield trees that unite cycads and angiosperms in a clade that is sister to a clade in which Gingko and Cupressophyta are successive sister taxa to gnetophytes plus Pinaceae. Application of a cycads + angiosperms backbone constraint in analyses of a morphological dataset yields better resolved trees than do analyses in which extant gymnosperms are forced to be monophyletic. The results have implications both for our assessment of uncertainty in trees from sequence data and for our use of molecular constraints as a way to integrate insights from morphological and molecular evidence.

Figure 1.

Putative phytochrome gene phylogeny that supports the orthology of gymnosperm PHYN, PHYO and PHYP (grey shading) with angiosperm PHYA, PHYC and PHYB, respectively, and that supports the monophyly of extant gymnosperms.

Figure 2.

Optimal gene phylogeny for angiosperm and gymnosperm phytochromes that was inferred in ML analyses of the 380-sequence phytochrome amino acid matrix using the PHY-specific amino acid rate matrix. The –ln L = 84272.061. Branch support is shown for selected nodes. The monophyly of PHYA and PHYC is supported by bootstrap values of 100%. A detailed tree is available in the electronic supplementary material, figure S1.

Figure 3.

The optimal PHY tree in figure 2 can be folded into a species tree without invoking undetected duplications or losses.

Figure 4.

Strict consensus of 20 trees of 346 steps that were inferred in the MP analyses of the Doyle (2008) morphological dataset. The optimal PHY topology was enforced as a backbone constraint. Bootstrap values from constrained analysis (leftmost number) and posterior probabilities (rightmost number) from unconstrained analysis are on the backbone nodes.

Figure 5.

(a) Unrooted topology for the rooted tree inferred by Rai et al. (2008). (b) Rooted topology inferred by Rai et al. using ML. (c) Rooted topology inferred by Rai et al. using MP. (d–h) The five additional possible rooted topologies.