Starbeast3 node age calibration

[Else]

Hello,

I am running a Starbeast3 analysis and my goal is to understand the relative divergence dates of my study species. I don't have any estimates of their substitution rates and their fossil record is very poor, but I do have a single point estimate for the age of the root, so I would like to calibrate the ages of the other nodes relative to that node.

Is it appropriate to estimate the phylogeny with branch lengths in substitutions/site, and then simply convert the branch lengths to years by scaling it according to the age of the root, or could there be any issues if I don't include the date calibration within the BEAST analysis so that it estimates branch lengths in years? i.e., is it necessary to provide BEAST with a MRCA prior and estimate the clock rate, or can I run BEAST without any information about the time calibration and just rescale the resulting tree after it is done?

Thank you,

Else

[Jordan Douglas]

Hi Else,

I would recommend you do the former (ie. provide BEAST with an MRCA prior and estimate the clock rate). This way the uncertainty of your divergence times will be captured by the posterior distribution. If you rescale divergence times after doing the Bayesian analysis, then the posterior distribution will be different, and no longer describing the same generative process of the model. 

Jordan