HOW TO ESTIMATE TRUE DIVERGENCE TIMES OF EXTANT LINEAGES USING THE AGES OF THEIR OLDEST FOSSILS AND AN ULTRAMETRIC MOLECULAR PHYLOGENY

When using the fossil record to determine the time of origin of a lineage, one is confronted with the fact that the oldest known fossil of a lineage will post-date its true time of origin. One way to overcome this problem is to add a confidence interval to the base of the lineage's observed stratigraphic range. However, computing the confidence interval requires knowing the density of fossils over the lineage's observed stratigraphic range, but this information is not readily available. Here I show how to bracket the true divergence times of the lineages in a molecular phylogeny if one has the age of the oldest fossils for at least two lineages and if the molecular phylogeny was derived from DNA sequences that evolved in a clock-like fashion (i.e., from DNA data that are ultrametric). The method first identifies the lineage that has the greatest proportion of its true temporal range captured by the fossil record. This calibration lineage is simply the lineage with the slowest estimated rate of molecular evolution. I then show how the ages of the oldest fossils from the remaining lineages, in conjunction with their respective branch lengths on the molecular phylogeny, may be used to place a confidence interval on the true time of origin of the calibration lineage, thus making it possible to bracket the true divergence times of all the lineages on the phylogeny. While most molecular data are not ultrametric, methods are being developed for converting non-ultrametric phylogenies into ultrametric ones (e.g., the program r8s). Thus, the method developed here may be broadly applicable. The method assumes that fossilization is random and is based on the conservative assumption that each lineage is represented by a single fossil. The impact of violations of these assumptions is assessed. The method may also be used to identify oldest fossils that have been grossly mis-dated or mis-assigned to the molecular phylogeny.