Q: PRIMARY KEY vs UNIQUE constraint

What is the benefit in defining the structure of a table as a PRIMARY-KEY vs a UNIQUE constraint in an Informix database? In other words, (I have looked in the manuals but could not find anything), how does Informix treat the PRIMARY-KEY differently than a UNIQUE constraint? They both use a unique index and seem to be different only in the table sysconstraint's constraint type (I forget the column name).