Assessing the effect of time-scaling methods on phylogeny-based analyses in the fossil record

Phylogeny-based approaches can be used to infer diversification dynamics and the rate and pattern of trait change. Applying these analyses to fossil data often requires time-scaling a cladogram of morphotaxon relationships. Although several time-scaling methods have been developed for this purpose, the incomplete sampling of the fossil record can distort the apparent timing of branching. It is unclear how well different time-scaling methods reconstruct the true temporal relationships or how any such inaccuracy could affect tree-based evolutionary analyses. I developed process-based simulations of the fossil record that allow the comparison of approximated time-scaled trees to true time-scaled trees. I used this simulation framework to test the effect of time-scaling methods on the fidelity of several commonly applied tree-based analyses, across a range of simulation conditions. When the fidelity of time-scaling methods differed, the stochastic cal3 time-scaling method with ancestral assignment produced preferable results. Estimating rates and models of continuous trait evolution was particularly sensitive to bias from scenarios that forced the insertion of many short branch lengths, a bias that is not solved by any of the considered time-scaling methods in all scenarios. The cal3 method of time-scaling can be recommended as the preferred time-scaling method among those tested, but caution must be exercised because tree-based analyses are prone to easily overlooked biases.