The joint prior density of divergence times, conditional on root age, and other time constraints are derived from the BDS model.
Consider a tree with The program BEAST implements special cases of the BDS model to generate a time prior (Drummond and Rambaut 2007).
A global clock model may appear sensible in these circumstances because it involves fewer parameters and because of lower rate variation due to shorter time scales and similarity of taxa. The first of these is the recursive procedure implemented in the program MULTIDIVTIME (Thorne et al. Time constraints are represented as “hard” maximal and/or minimal bounds on node ages (sensu Yang and Rannala 2006).
Nevertheless, a relaxed clock is often used, possibly because it accommodates both homogenous and heterogenous rates (e.g., Warren et al. The second method for specifying this prior on times uses the birth–death process with species sampling (BDS).
This is implemented in the program MCMCTREE (Rannala and Yang 2007; Yang and Rannala 2006; Yang 2007).
The BDS prior is specified by 3 parameters: per-lineage birth (λ) and death (μ) rates and the sampling proportion (ρ).
Widths of posterior intervals were also very dependent on the position of the calibrated node within the tree, particularly when sequences were short.Divergence events since the mid–late Miocene generally correspond to lower taxonomic levels, such as the genesis of new intrageneric or intraspecific lineages (Avise 2000). Simulations indicate that global clock analyses perform relatively poorly when the clock is violated but relaxed clock analyses can work well for clock-like trees (Ho et al. Little is known about the sequence information content required for accurate recovery of divergence times.Knowledge of their timing can provide valuable insights into the climatic and physical processes that have shaped genetic diversity within species as well as into speciation events. Posterior estimates of divergence times will reflect the prior when sequences are short (and therefore less informative) but should be increasingly influenced by the likelihood as sequence length is increased.These include the Yule process (lineage birth only, μ = 0, ρ = 0) and birth–death models (lineage birth and death only, ρ = 0).
BEAST assigns hyperpriors on λ and μ, whereas all BDS parameters are specified by the user in MCMCTREE.Bayesian methods are increasingly being used to estimate divergence times without the restrictive assumption of a global clock.