Phenetic and molecular phylogenetic data can be noncongruent for several reasons and that is a quite normal situation. However, you can try to use supertree algorithms which allow creating one phylogenetic tree from several ones even if they were built by different methods. Another possible way to solve this problem is using Bayesian inference for all data if it is possible in your research.

Sometimes maximum parsimony is good way but my colleague above is right that Bayesian approach is good way to solve your problem. Nonetheless ITS sequences are source of problems if you do not know nothing about their mutation rate. Congruences may be very difficult to find if the selective pressure has been significant. It is good idea to use another set of sequences from nuclear or mitochondrial DNA.

thanks all, that was a nice advice and help me to solved my problem. I will start from study Bayesian approach, it would be nice.

Thanks again!