Molecular clock estimates of the age of Angiosperms are generally much older than fossil evidence suggests and therefore are controversial. Here we explore the effects of DNA substitution rate heterogeneity on molecular clock estimates using a data set of nine genes from the chloroplast (atpB, rbcL, matK), mitochondrial (atp1, matR, and SSU and LSU rDNA), and nuclear genomes (18S and 26S rDNAs). A global molecular clock can be rejected for all genes, and all combinations of genes, under all models of evolution explored. This is due in part to an increase in absolute substitution rate in 3-4 lineages, including monocots, Piperales, Ceratophyllum, and Ranunculales in the eudicots. This increase in rate is found in all three plant genomes. Alternative analytical methods have been proposed to estimate divergence times in the absence of rate constancy. We explore three of these methods to estimate the age of angiosperms. For each method, we also examine the effects of using different calibration points and different tree topologies. The choice of calibration point has an enormous effect on the estimated ages. In contrast, tree topology appears to have little effect.

Key words: Angiosperm phylogeny, molecular clock, rate heterogeneity