Ancestral states reconstruction for polymorphic taxa in Mesquite will work only under parsimony reconstruction.

Hi Luiz,

Yes, I noticed this, but now I know that if you transform the polymorphic taxa into the third (or fourth) state, then it works. Thanks for your kind reply!

Hi Xiangqin,

In case of polymorphic characters, it may be better for the accuracy of your ancestral state reconstructions to code polymorphisms as missing data instead of a distinct character state. When polymorphisms are coded as missing data, states for the internal nodes are reconstructed considering all possible states for that character...and later you can make your own conclusion whether it's likely that polymorphism was present at each particular ancestral node based on the proportional likelihood scores.

In general, using polymorphic characters in a phylogenetic analysis is a questionable matter and you may find very different opinions on this topic. Here's one of the many papers that may be of an interest to you if you wish to look into this topic a bit more:

Kornet and Turner. 1999. Coding Polymorphism for Phylogeny Reconstruction. Syst. Biol. 48(2):365–379 (pdf copy attached)

Cheers,

Oksana

Hi Oksana,

Thanks for your reply and the attached reference. I think if a species was coded as missing data, then it seems that we will have it excluded from the analysis, that is really the situation I do not want.

Hi Xiangqin,

I agree with Oksana that replacing polymorphic entries by missing ones can be a solution for your issue since missing entries will be analysed as any possible state for that character. This procedure will probably cause no harm to your analysis for binary characters but this is not necessarily true for multistate characters when one or more states are actually absent in a polymorphic taxon.

Another solution could be splitting each polymorphic taxon into monomorphic subtaxa, as suggested by Nixon & Davis (but also see Watanabe's arguments favouring state frequencies when available). Papers follow attached.

Cheers

Luiz

Hi Luiz,

Thanks for the attached references. Yes, when we have multiple states, then the solution that Oksana proposed will be problematic. If we have 3 or more states, we could treat the third (fourth) as another separate state, for example, if we have 3 states (0, 1, 2), the taxa with 0 & 1 could be treated as 3. This is exactly the method used in the attached paper published in 2012.

Best,

Xiangqin