Classification

Classification is the complement of ordination (Chap. 4 ): while ordination summarizes continuous trends in a multivariate data set, classification seeks discrete groups. With species data sets, these groups are community types; with environmental variables, the groups are biophysical settings or habitat types. As with ordination, there are many alternative tools available, with a long history in ecology. Here we begin with perhaps the most common approach, cluster analysis. The workflow begins with exploratory data analysis, then proceeds to post-processing to interpret and communicate the results. This post-processing might summarize the groups in terms of the variables used to define them (e.g., indicator species for compositional types, environmental predictors for habitat types) or it might crosswalk the data (e.g., indicator species for habitat types or environmental predictors for community types). As with ordination, classification provides a powerful summary for the presentation of data sets collected as part of an inventory or monitoring program. Ordination and classification are especially powerful when integrated together.