Can also quantify how representative a particular location is of its parent ecoregion

The final centroid of each cluster, since it represents the arithmetic average of all member cells, is the best single way to represent that cluster ecoregion

Map cells which are in the interior of the cluster, close to the mean centroid, are highly representative of this ecoregion - Map cells in the outer "shell" of the cluster are less representative

The Euclidean distance from each cell to it's assigned centroid is an easily quantifiable measure of representativeness for that location

Such a regional representativeness value can be computed for all locations in the map

Takes into account the level of ecoregion division - Always relative to current ecoregion's centroid

Hargrove, Hoffman (1999) Using multivariate clustering to characterize ecoregion borders. Computers in Science and Engineering 1(4):18-25.