[Taxacom] minimal estimates

[Sergio Vargas]

  Hi all,

I just want to point that most molecular clock estimates rely on a 
maximum age for the root. Without this strong assumption (i.e. a maximum 
age for the root cannot be justified in most cases, I would say in all 
cases) the method just doesn't work.

Is also important the way in which you model the rates. I've seen nodes 
showing outliers going back 2000 million years back in time. Most of the 
time you don't detect this outliers because they are not included in the 
95% CI of the mean root age reported by the program. Anyways, the 
analysis is generally hard from the beginning (establishing the fossil 
calibration) to the end (interpreting the dates), and most researchers 
don't like to acknowledge that the dates are conditional on the root age 
used.

sergio


> John Grehan and co-thinkers point to a very important limitation of molecular dating exercises - that the first appearance of a clade in the fossil record provides only a minimal age for that clade.
>
> However, it does not follow from this that molecular clock dates for clades calibrated on the basis of such fossils are necessarily minimum estimates.
>
> Why? Because there are many potential sources of error to be considered. Some of these, such as misspecified substitution models, ancestral sequence polymorphism, incorrectly estimated phylogenies, hemiplasy, and among lineage rate variation can all result in over-estimates of clade ages. If errors leading to age-estimate exaggeration are greater than any fossil record related error causing age underestimation, then the age-estimate be to old.
>
> It is not helpful to seize on one source of error alone, we need to consider all the potential sources of error, and preferably quantify them.
>
> Not all of these estimates are of equal reliability, and in general I do agree with John that they need to be treated with scepticism. However, to label them minimum estimates is incorrect, even if it does tend to favour panbiogeographer's eccentric ideas about the age of taxa.
>
> Cheers
> Rob