Another look at the root of the angiosperms reveals a familiar tale

Since the advent of molecular phylogenetics more than 25 years ago, a major goal of plant systematists has been to discern the root of the angiosperms. Although most studies indicate that Amborella trichopoda is sister to all remaining extant flowering plants, support for this position has varied with respect to both the sequence data sets and analyses employed. Recently, Goremykin et al. (2013) questioned the "Amborella-sister hypothesis" using a "noise-reduction" approach and reported a topology with Amborella + Nymphaeales (water lilies) sister to all remaining angiosperms. Through a series of analyses of both plastid genomes and mitochondrial genes, we continue to find mostly strong support for the Amborella-sister hypothesis and offer a rebuttal of Goremykin et al. (2013). The major tenet of Goremykin et al. is that the Amborella-sister position is determined by noisy data--that is, characters with high rates of change and lacking true phylogenetic signal. To investigate the signal in these noisy data further, we analyzed the discarded characters from their noise-reduced alignments. We recovered a tree identical to that of the currently accepted angiosperm framework, including the position of Amborella as sister to all other angiosperms, as well as all other major clades. Thus, the signal in the "noisy" data is consistent with that of our complete data sets--arguing against the use of their noise-reduction approach. We also determined that one of the alignments presented by Goremykin et al. yields results at odds with their central claim--their data set actually supports Amborella as sister to all other angiosperms, as do larger plastid data sets we present here that possess more complete taxon sampling both within the monocots and for angiosperms in general. Previous unpartitioned, multilocus analyses of mitochondrial DNA (mtDNA) data have provided the strongest support for Amborella + Nymphaeales as sister to other angiosperms. However, our analysis of third codon positions from mtDNA sequence data also supports the Amborella-sister hypothesis. Finally, we challenge the conclusion of Goremykin et al. that the first flowering plants were aquatic and herbaceous, reasserting that even if Amborella + water lilies, or water lilies alone, are sister to the rest of the angiosperms, the earliest angiosperms were not necessarily aquatic and/or herbaceous.