Plant evolutionary genomic analyses have the greatest potential to reveal phylogeny when congruent evidence is obtained independently from the plastid, mitochondrial, and nuclear genomes with all methods of analysis. Incongruent genomic compartment estimates were obtained in analyses of previously assembled multigene data sets for basal angiosperms due in part to a combination of suboptimal outgroup choice, phylogenetic noise, and incorrect homology assignment. In spite of these potential sources of systematic error, method of analysis had the strongest effect on the results obtained. In this study, results are presented from combined analyses of 3 mitochondrial (cox1, atpA, matR), 2 plastid (rbcL, atpB) and 1 nuclear gene (18S) that suggest Amborella+ Nymphaeales as the first-branching angiosperm lineage. Individual genomic compartments and most methods of analysis estimated the Amborella+Nymphaeales topology with high support. Surprisingly, unweighted parsimony alone estimated Amborella-only as the first branching angiosperm, but support for this result was low after noise reduction. Ancestral character state reconstructions differ between the two topologies and profoundly affect inferences about angiosperm evolution. Relative Apparent Synapomorphy Analysis (RASA)-based data exploration was a critical step involved in these analyses and illustrated the advantage of evaluating data prior to formal phylogenetic study.

Key words: Amborella, atpA, cox1, evolutionary genomics, Nymphaeales, RASA