Case study 7 - Biogeography of late Ordovician brachiopods


Data file: hirnanti.dat

During the latest Ordovician (Hirnantian), three discrete brachiopod provinces have been identified, developed against a background of regression associated with a major glacial event. The typical Hirnantia fauna (Kosov province) occupied temperate and subtropical latitudes whilst the Edgewood province straddled the equator and the atypical Hirnantia fauna (Bani province) circumscribed the south pole:

Rong & Harper (1988) recorded the distribution of 42 genera across 32 Hirnantian localities. The modified data set is available on hirnanti.dat.

Jaccard and Dice coefficients

Open the file Hirnanti.dat, select all, and choose 'Dice/Jaccard' in the Statistics menu:

Note the sites with the highest mutual correlations.

More information about Jaccard and Dice coefficients can be found in the manual.

Cluster analysis

Clustering is carried out by rows. We want to cluster localities, which are currently in columns, so you must first transpose the matrix (Edit menu). Then select 'Cluster' in the Multivar menu. Try different algorithms and distance measures, and compare the results. Below is shown a dendrogram based on the Paired Group clustering of the Dice coefficients. Can the three provincial groupings established by Rong & Harper (1988) be confirmed?

More information about the clustering module can be found in the manual.

Seriation

For seriation, the taxa need to be placed in columns again. Select 'Transpose' in the Edit menu if necessary. Then choose 'Seriation' in the Multivar menu. The calculation time may be substantial. Considerable opportunity is available to experiment with constrained and unconstrained seriation on this large data set. Below is shown the result of an unconstrained seriation. Can the starting matrix be transformed to approximate to the faunal gradients already established qualitatively for the Hirnantian?

More information about the seriation module can be found in the manual.

Suggested answers

Next: Case study 8