Hello Yasser,

It kind of depends. For example, if you have NGS data with thousands of loci then 0.70 is not well-supported at all since more data often inflates support and if you can not get

Hello Yasser,

It kind of depends. For example, if you have NGS data with thousands of loci then 0.70 is not well-supported at all since more data often inflates support and if you can not get

Max is right that it is a matter of opinion on what value for a posterior probability is recommended to feel confident that a clade is well supported. This is a very subjective judgement.

In statistics we are only ready to consider that two values are significantly different from one another p-value is less than 0.05 or more conservatively if its is less than 0.01. These are arbitrary values but are often thought to provide an acceptable measure of confidence in the result. 

The corresponding confidence measures in phylogenetics are posterior probabilities and bootstrap and aLRTS. Values of probability of 0.95 or 0.99 are considered really strong evidence for monoplyly of a clade.

However in the case of a phylogenetic tree, each branch is really a separate hypothesis so that the problem of multiple tests on the same set of data.So these measure of confidence are vastly biased upwards. Although rarely done, these values should be Bonferroni (say) corrected for multiple tests.

Many researchers consider values above 0.80 as good evidence for the monopoly of a clade even though the corrected confidence value in that clade us far lower than 0,80

The bottom line is that the a really well supported tree should have uniformly high values of posterior probabilities or bootstrap values. My cutoff is 0.90 and I feel much better with values greater than that.

Please also remember that if your model of evolution is unrealistic, high PPs, aLRTs and Bootstrap values will be very misleading.