Two-way indicator species analysis - TWINSPAN
|
|
Top Previous Next |
|
TWINSPAN is somewhat complex divisive clustering method originally devised by Hill for vegetation analysis, but quite suitable for animal communities as well. An interesting feature of TWINSPAN is that it forms what are termed pseudospecies. These are separate variables for the different levels of abundance of a species. Samples are ordinated using Reciprocal Averaging (RA). A dichotomy is then made using the RA centroid line to divide the samples into two groups (negative and positive). This dichotomy is then refined using an iterative procedure. The clusters of samples obtained are then ordered so that similar clusters are near each other. This procedure continues in a hierarchical fashion to subdivide the groups until the minimum group size initially selected by the user is obtained. Species are then classified using the sample (quadrat) classification. In the original output a table is then produced showing species-by-site (quadrat or sample) relationships.
Once the TWINSPAN option is selected from the drop-down menu you will be presented with the Setup for TWINSPAN window in which options can be selected. If no changes are made prior to clicking OK, then the default settings will be used. For many vegetational studies the defaults are appropriate.
See also Maximum size of the data set. |