enter image description here

I have a question about bootstrapping:

The way that I understand it is that each number represents how well a given node is supported. So in the diagram presented, we can see that the "Seal-Bear-Cat" clade is fairly well supported with a bootstrap value of 87. From that, the "Seal-Bear" clade has a bootstrap value of 95, making it quite well supported within the larger clade.

My question is this: Does that 95 for the "Seal-Bear" clade refer to the idea that that given relationship is a true 95, and therefore very well supported, or is it a value of 95 within 87, making it a true value of somewhere closer to 83 for the overall tree?

  • $\begingroup$ Welcome to Biology.SE. Your title could more accurate (and then attract more users). It is typically pointless to use the word "question" in the title given that obviously the post contains a question. $\endgroup$ – Remi.b Oct 13 '15 at 15:11
  • 3
    $\begingroup$ The term bootstrapping is a concept of statistics that is much broader than its particular application to phylogenetics. Bootstrapping refers to the process of sampling iteratively your own sample in order to approximate the sampling distribution and calculate confidence intervals. $\endgroup$ – Remi.b Oct 13 '15 at 15:13
  • 1
    $\begingroup$ Bootstrap support is merely the percent of bootstrap iterations that support the clade, it doesn't account for tree hierarchy. $\endgroup$ – Eli Korvigo Oct 13 '15 at 16:51
  • $\begingroup$ where is this slide from? Could you give us some references? Bootstrapping can be used in many contexts as @Remi.b pointed out. $\endgroup$ – Memming Oct 13 '15 at 23:06
  • 1
    $\begingroup$ I've voted to close the question unless more contexts can be given. $\endgroup$ – SmallChess Oct 14 '15 at 1:12

If the values on the branches indeed represent bootstrap supports, they indicate the percentage of times a clade has been obtained in a set of trees computed from resamplings of a starting data matrix. This is irrespective of the topology within the clade or outside it, because this is merely a count of bipartitions (a branch corresponds to a bipartition of the taxa in the tree).

I think there is no real sense in multiplying bootstrap percentages along a tree, contrary to what the end of your question suggests.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.