# Branches created in the Explorer tree

The first branch takes the square root of the full matrix <ins>Frierfjord macrofauna counts</ins>, giving <ins>Data1</ins>, on which sample Bray-Curtis is calculated, <ins>Resem1</ins>. This was only used to seriate the *x* axis on the original shade plot but, as seen above, **Special>Reorder**>Samples>(Order•Original) in place of the default (Order•Seriate) has restored the axis to the label order of the data matrix. If the **Wizards>Matrix display** default of <u>not</u> retaining sample groups had been followed (no factor supplied), then <ins>Resem1</ins> would be input to **Analyse>Cluster>CLUSTER**, creating  a dendrogram (without running  SIMPROF), displayed on the *x* axis and with <ins>Resem1</ins> used to seriate samples within the constraints of dendrogram rotation. <ins>Resem1</ins> is the right resemblance matrix to use for multivariate routines such as *n*MDS and ANOSIM. The second branch starts with a **Tools>Duplicate** copy (<ins>Data2</ins>) of <ins>Frierfjord macrofauna counts</ins> on which **Select>Variables**>(•Use n-most important where n is <ins>50</ins>) has been run. It is species-standardised by **Pre-treatment>Standardise**>(Standardise•Variables) & (By•Total) to give <ins>Data3</ins>, on which **Analyse>Resemblance**>(Measure•Index of association) & (Analyse between•Variables) then gives the species similarities <ins>Resem2</ins> on which CLUSTER is run in just the same way as it would be for samples. \[The Standardise step is not really needed here because IA will restandardise species again as part of its equation. It is included partly to remind you that there is a species standardisation step but also because there are other cases, such as the Type 3 SIMPROF tests for *coherent species curves* (statistically distinguishable species clusters) later in this section, in which an initial species standardisation is required <u>even though</u> an index of association will be calculated afterwards, so this is a good habit to adopt. (The issue arises there because the permutation direction in Type 3 SIMPROF is <u>across</u> species, and this only makes sense if species are scaled to add to the same total).\] The final sub-branch in the Explorer tree, off the data matrix <ins>Data2</ins>, with its reduced number of species, is the one that generates the Shade Plot. <ins>Data2</ins> is transformed with the specified square root, to give <ins>Data4</ins>, which is input to **Plots>Shade Plot** to give <ins>Graph2</ins>. If you repeat that last step manually, you will see that the resulting graph is a simple snapshot of the data matrix with samples and species in exactly the same order as the input matrix and no clustering or other ordering of the axes.