BIRCH

Workflow example: Analysis of population structure for green foxtail in Manitoba


Green Foxtail is a weed that grows widely throughout Manitoba. To find out whether the plants sampled across the province represent a single interbreeding population, of several distinct subpopulations, RAPD markers were screened in 12 Manitoba populations, plus the Chinese accession 439-86. This tutorial describes a simple workflow that uses parsimony analysis of these markers to assess population structure.

The first step is to read in Phylip-formatted marker file. Save the file manitoba.phyl  in your current working directory. In GDE, choose File --> Import Phylip Discrete Data, and type in the name manitoba.phyl.

The markers will be read into GDE. Next, select all markers and choose Phylogeny --> RFLP/RAPD/AFLP etc. Parsimony.


Set a random number seed for jumbling the order in which foxtail accessions are added to the tree, and click on OK. Output will be sent to several windows:

DOLLOP report in a text edit window (dollop_report.txt)

Phylip-formatted trees in a text edit window (dollop_trees.treefile)


Phylogenetic trees in a tGDE window


You will note that this search produced 9 equally-parsimonious trees. (The number of parsimonous trees can vary greatly depending on the input order of accessions. It is usually best to jumble the order several times, rather than once, as shown in this example).

The next step is to create a consensus tree from these 9 trees. In the tGDE window, select all trees, and choose Evaluate --> Consense.


No defaults need to be changed, so click on OK. Output will be sent to several windows.

Consense report (consense_report.txt)


The consensus tree (consense.treefile)


Consensus tree in the ATV tree editor


Consensus tree in tGDE


It is sometimes also instructive to view the tree in other types of displays. The Phylip Drawtree program provides draws radial trees. In the tGDE window, select the consensus tree and choose Draw --> Drawtree.



Click on OK to launch the program. By default, output will go to a PostScript viewer.


The results presented here are not by themselves compelling. Four of the populations (UM8, BDAY, G17 and R2G1) seem to cluster together, whilc no othe obvious clusters are apparent. As expected, the Chinese accession 439-86 does indeed behave as an outgroup. It would be best to re-do the analysis using bootstrapping to see how robust these groups are.