If you stay within one single peak type as in this case Mn2P3/2, the stoichiometric fractions are directly proportional to the peak integrals. Thus, if you want percentages for your charge states, you just have to normalize the sum of the peak integrals to 100%.
Side question: how large is your Mn3+ fraction? That's a rather labile species and therefore having a large fraction of it means you have to be really certain about your data.