FIGURE 1. Cluster selection and variance across clusters. These show that the difference between clusters and the total variance of each cluster declines with increasing numbers of clusters, but the gain is negligible for values greater than k = 40. 1.1) Scree plot of the Euclidean variance across cluster centroids. The red whiskers show that the total variance diminishes with increasing numbers of clusters, while the mean variance among the cluster centroids, indicated by the black dot, approaches zero. 1.2) Blue whiskers and black dots indicate that as the number of clusters increases, the mean beta diversity (difference) and its variance increase– clusters become more specific, and thus more different.