groups of observations, and cluster centroids
defined by the mean values are used to describe
each group. Proponents of this empirical
approach claim that it brings no preconception
to the mapping task but that it does emphasize
methods of numerical taxonomy in the search
for pattern recognition. Clustering algorithms
have been used (c.f. Hargrove and Luxmore
1998; Host et al. 1996; Mackey et al. 2008;
Snelder et al. 2009; Weigelt et al. 2013) to
produce ecological regionalization based on
environmental factors believed to control ecosystem patterns. They have been criticized as
inefficient (Rowe and Sheard 1981) and some
argue that boundaries of ecoregions cannot be
derived mathematically from one or more
datasets without understanding ecosystem pattern and process (Bailey 2004).
152
Appendix D: Comparison of Ecoregion and Related Approaches