Dating the Dipsacales: comparing models, genes, and evolutionary implications.

Dipsacales is an asterid angiosperm clade of ca. 1100 species, with most of its lineages occupying temperate regions of the Northern Hemisphere. A recent phylogenetic analysis based on 7593 nucleotides of chloroplast DNA recovered a well-resolved and strongly supported phylogenetic hypothesis, which we use here to estimate divergence times within the group. A molecular clock is strongly rejected, regardless of data partition. We used recently proposed methods that relax the assumption of rate constancy among lineages (local clocks, nonparametric rate smoothing, penalized likelihood, and Bayesian relaxed clock) to estimate the ages of major lineages. Age estimates for Dipsacales varied widely among markers and codon positions, and depended on the fossils used for calibration and method of analysis. Some methods yielded dates for the Dipsacales diversification that appear to be too old (prior to the presumed 125 my [million years] age of eudicots), and others suggested ages that are too young based on well-documented Dipsacales fossils. Concordant penalized likelihood and Bayesian studies imply that Dipsacales originated in the Cretaceous, as did its two major lineages, Adoxaceae and Caprifoliaceae. However, diversification of crown Adoxaceae and Caprifoliaceae mainly occurred in the Tertiary, with the origin of major lineages within these clades mainly occurring during the Eocene. Another round of diversification appears to have occurred in the Miocene. Several radiations, such as Valerianaceae in South America and Dipsacaceae around the Mediterranean, are even more recent. This study demonstrates the wide range of divergence times that can be obtained using different methods and data sets, and cautions against reliance on age estimates based on only a single gene or methodology. Despite this variance, significant conclusions can be made about the timing of Dipsacales evolution.