Dear Theo.

This can be done with the function rerootingMethod as described here, although it is *very* important to note that rerootingMethod is only valid if the fitted transition model is symmetric (e.g., "ER", "SYM", or any custom model in which i->j == j->i for all i & j). It can also be done with make.simmap for both symmetric & non-symmetric transition models. If you sample enough stochastic maps & then run summary on the set of maps the posterior probabilities will converge on the marginal ancestral states under the defaults for make.simmap, so long as a flat prior is used on internal nodes. This is also on my blog here:

All the best, Liam

Liam J. Revell, Associate Professor of Biology
University of Massachusetts Boston
& Profesor Asociado, Programa de Biología
Universidad del Rosario

On 6/21/2018 5:32 PM, Théo Léger wrote:

I am working on the phylogeny of a subfamily of moths and I use ace from the 
ape-package to reconstruct the ancestral state of a bunch of morphological 

I encountered a problem with the few unknown states I have on my matrix (coded 
"?", either because material for examination was missing or the state could not 
fit in any of the categories): they are treated as other characters. Is there a way to 
ignore them? Is there a way to estimate the state for the species with missing 
information? I know it is possible to estimate the state at a tip with missing 
information in functions like AncTresh (using bayesian reconstruction) from the phytools 
package, but I am not sure if ML reconstruction can do it.

Thanks in advance to those who can enlighten me!


Th�o L�ger

        [[alternative HTML version deleted]]

R-sig-phylo mailing list -
Searchable archive at

R-sig-phylo mailing list -
Searchable archive at

Reply via email to