As Yuri mentioned you can use TNT for a parsimony approach or you can use MrBayes for a Bayesian approach. These are probably the easiest and most robust methods.
You should apply analysed under maximum parsimony, followed by successive reweighting and implied weighting as optimality criteria. Tree searches by using a Wagner tree as the starting tree plus tree bisection and reconnection (TBR), with ten trees saved per replication and also by TBR branch swapping. Analyses with implied weighting were conducted in TNT 1.1
(Goloboff, Farris & Nixon, 2008) Goloboff
et al. (2008). you should used maximum parsimony with successive reweighting by PAUP 4.0b10 (Swofford, 2002). Character supports calculation with TNT.
Actually now you can combine PAUP or TNT to the mesquite itself using a package called Zephyr and it makes the whole parsimony approach much more easy.