Classification of Reference Sites

As the aim is to predict the expected biological communities for sites, the next step is to classify the reference sites into site groups based solely on their biological composition. Sites with similar communities of macroinvertebrates are placed in similar groups. Several different multivariate methods of classifying the reference sites exist eg unweighted pair-group mean arithmetic averaging (UPGMA). The best classification method will be the one which leads to the most accurate prediction of indices of quality for the reference sites. In the UK, the method called TWINSPAN (Two-way indicator species analysis) was considered the most appropriate. TWINSPAN has been used successfully many times in various branches of community ecology. It repeatedly divides the reference sites into sub-groups in a hierarchical manner. At each stage, the reference sites in a particular sub-group and their taxa are simultaneously ordered so that sites with similar taxa are placed close together on the ordination axis. The position of sites along the axis is used to split the group in two. This classification involves information on both presence-absence of species and family level abundances.

The initial classification of the 614 reference sites for mainland Britain split into four groups

Most importantly, it must be emphasised that the biological variation across the region is a continuum and that sites do not naturally fall into completely distinct biological types. The biological classification of reference sites into groups is just an intermediate convenience and not a definitive assignment of sites to particular individual groups.