Dear Viji

Maybe this answer helps you...

https://www.researchgate.net/post/hi_can_someone_tell_me_the_relationship_between_SNPs_number_and_quality_of_a_phylogenetic_tree

greetings

carolina

The values are simply the percentage of bootstrap replicated datasets that produce a tree on the bootstrap tree. Some branches in your tree are very well supported by the sequence alignment you are testing and those are the ones with the 100% bootstrap value. Values such as 10, 28 and 47 show that the sequence data fails to support those particular branches. These show up in your neighbor joining tree but the confidence in those branches is extremely low given the data.

Some people think that anything over about 80% is good support for a branch (or a clade). Because of something called multiple tests bias, I prefer to trust only those values that are above 95% at the least.

Depending on what you are trying to demonstrate, it may not matter what most of the branches show. If your interest is primarily in the ones with very good bootstrap proportions then the analysis may be quite adequate. In other words demonstrating monophyly for a clade of interest may satisfy your hypothesis regardless of what the rest of the tree shows.

An important caveat here is that any bootstrap value can be positively misleading if the assumptions of the model and the analysis are incorrect. A NJ tree has some rather bad assumptions that could easily give you confidence in the wrong tree.

That is why maximum likelihood with realistic assumption (a model that fits your data) is generally very much favored over a simple algorithmic method like UPGMA or Neighbor Joining. ML analysis is able to incorporate many of the processes that seem to govern the evolutionary history of a set of sequences.

Let me know please if this is not clear to you.