The target tables in the first distribution (Oracle_new) are populated with snapshot copies of the source table columns selected for distribution. The result at the target that used the reference mapping is the same as the result at the target it referenced.
The target tables in the second distribution (OraMssMerge) have a mixture of source and target column values in the records whose keys matched copied source records. In addition, they have some entirely new records from the source (new, unmatched keys) and some of the same records (keys not matched in the source copy) they had before.