PiscesLogoSmallerStill Worked Example - Stream Invertebrates

Top  Previous  Next


To help you understand the full capabilities of the program and its methods, we present below a worked example, using one of the data files installed with the program: River inverts.csv.


The data were collected in 2000 as part of an investigation into the effects of river restoration on biodiversity in several chalk streams in southern England. The report on the investigation, Effects of physical restructuring of channels on the flora and fauna of three Wessex rivers, can be downloaded in Adobe Acrobat PDF format from our website at consult.pisces-conservation.com/latestreports.html


Concentrating on macroinvertebrates, the data comprise percentage composition by family from restored and unrestored reaches of the 3 rivers and 2 tributaries. There are 61 families (rows) across 49 sample sites (columns).


The site names are coded with the initial letter of the river, the site name, and 'U' or 'R' for unrestored or restored. The rivers are the Wylye, its tributary the Till, the Piddle, its tributary the Devil's Brook, and the River Avon. So for instance, the unrestored stretch at Hyams Farm on the Avon is coded Ahyam U, and the restored site at Great Wishford on the Wylye is coded Wgwish R.


To begin the example, open the data file C:\My Documents\CAP 5 Data\Demo Data\River_inverts.csv  from the File menu. On opening the file, a warning box pops up, to inform you that the row titled 'Hydrida' has no data - i.e. all the values are zero. The program will automatically remove this row from the Working Data set.


Let us first perform a Principal Component Analysis; from the Ordination menu, choose PCA - Covariance. Click on the PCA Plot tab which appears at the bottom, to view the resulting chart (shown below with all labels removed, for clarity). Note that, because the PCA depends on a randomisation of the data, the chart may appear as a mirror image of the one below. This is unimportant.





Immediately, we can see that most of the species vectors are tightly grouped in the centre of the plot, but there are 3 strong vectors and a number of sites associated with them. By experimenting with the labelling, we can see that the three families are Simuliidae, Hydrobiidae and Gammaridae. By looking at the site labels, it becomes obvious that the 8 sites plotted in a line to the right are the A sites, from the River Avon. The vector suggests that these sites are strongly correlated with the Hydrobiidae. Similarly, the site at the top left, Wyarn U, is strongly correlated with the Simuliidae. We can confirm this from the working data; over 90% of the individuals in the Wyarn U sample were Simuliidae.


To  give the chart more impact, we should now label the chart to show the vector titles etc; follow the instructions for adding annotations under Preparing Charts for Output.




Similar conclusions can be reached by performing Reciprocal Averaging; choose Ordination : Reciprocal Averaging to perform the analysis. After adding annotations and lines, the plot appears as below:





W Yarn U is correlated with the Simuliidae and Ptychopteridae, whereas the group of 8 Avon sites is strongly associated with Hydrobiidae, Tabanidae and Leptophlebiidae.