Competitive hierarchies, reversals of rank order and the de Wit approach : are they compatible?