When the competition between two species needs to be tested, species need to be grown together. However, the specimen distribution within the experimental units (pots) is almost as important as the number of replicates that we are using (see Experimental design). The different species need to alternate and be evenly distributed in the experimental unit.
Once all treatments have been set up, the experimental units typically need to be randomly distributed in the available space. The reason behind distributing the experimental units randomly is to avoid the interference of any local condition on certain treatments (if they are allocated together). By doing this, all treatments will be equally affected by the conditions in the experimental area.
However, if the number of replicates is low (i.e. less than 5), it is more convenient to use a stratified random design. In a stratified random design, the replicates are first divided into groups and then distributed randomly among them. By doing this, it is ensured that the three replicates of the same treatment are not placed close to each other, and thus, spatial bias is avoided.