In Mrbayes, What Does It Mean For A Node To Have More Than Two Descendants?

Question

[Douglas S. Stones]

In MrBayes it's possible to generate phylograms that contain nodes with more than two descendants. Such an example arises in the MrBayes manual (see this) for the basic primates.nex.

Question: How should nodes with more than two descendants be interpreted?

Strictly speaking these should be bifurcation trees (i.e. each node has two descendants). My guess is that two speciation events occurred so closely together that MrBayes is unable to accurately determine which occurred first. If that is correct, I'd then appreciate a pointer to a document that can confirm this guess.

Answer 1

[David W]

Hi Douglas,

In phylogenetics multifurcations are usually called "polytomies" and they come in two varieties.

A 'hard' polytomy is just as you describe, a result of very rapid divergence making the order of branching almost impossible to recover. (Biologically speaking, these trees are often associated with adaptive radiation).

A soft polytomy occurs between quite distantly related taxa when we don't have enough information to infer the relationships between them. So, with MrBayes the most common topology makes up less than 50% of the the trees in an MCMC sample, and a consensus tree will have all the taxa in question descend from one node. (If you want to force it to be dichotomous you can use "all comptaible clades" as the consensus type.

I'm not sure about a general reference for this, but there is Sys Biol. paper on polytomies in Bayesian phylogenetics in particular.

[Douglas S. Stones]

Thanks, that's really helpful.

Answer 2

[Rvosa]

Are you finding the polytomies in the consensus result or in individual trees that were sampled during the run? If the former, the polytomies are simply because no possible resolution for that node had a frequency about the cutoff (e.g. 50%). If the latter, it would be because MrBayes is sampling unresolved topologies (as discussed by the article that David references, though I don't think MrBayes actually does that).

By the way, in the MrBayes wiki, I'm not seeing polytomies in the two trees at the bottom of the page. The root node is the plus sign, so Lemur catta and Tarsius syrichta do come out as a bifurcating clade, but perhaps I'm interpreting it wrongly or maybe you're referring to something else.