Permanova (vegan::adonis2) with two factors but one factor is significant for vegan::betadisper

[freixas84]

really appreciate your help on this. I have a data matrix of relative abundance of 20 taxa identified at the genus level via 16s. They belong to 100 human subjects. Subjects were randomly grouped into 5 treatments (20 subjects per treatment). Microbiome 16s was measured at 4 timepoints. So basically data can be organized as follows: 100 subjects, 5 treatment (20 subjects per treatment), 4 timepoints per treatment.

MY QUESTION: When running betadispers prior to PERMANOVA, I get a significant P-value for the treatment factor, but not for timepoints within treatment.

Should I run a Permanova for every treatment group independently given the fact that treatment factor does not meet the assumption for Permanova?

Any help/suggestion would be greatly appreciated.