On the need for manipulating density in competition experiments