ABSTRACT
The ordering of the environments on the horizontal axis can be chosen to be the same for each of the attributes to make it easier to compare responses across environments from one attribute to another. If the environment scores or indexes are reasonably distinct then they could provide both an ordering and a spacing on the horizontal axis. A relatively simple alternative is to use a greedy close ordering of the environments which results in having side-by-side those environments that are as little different as possible. The correlation coefficients amongst the environments provide a suitable similarity measure which has been standardized for the variability within each environment. The two environments which are most similar are placed closest together — linked first. The second link joins those environments which are next most alike.
