5  Phylogeny reconstruction and dating

5.1 Phylogenetic reconstruction and divergence dating in Beast

5.1.1 BEAUti

BEAUti is located in the BEAST software folder. Import the nexus file into software BEAUti.

5.1.2 Set up the Beast xml file.

Select the following options in BEAUti:

a. Site Model Tab: We’re going to use the HKY+I+G model (normally you would first find the best model using a software like jModeltest). Select Estimate substitution rate; set 4 Gamma rate categories; Estimate Shape (Leave at default 1.0); Proportion of Invariant sites: 0.1; Estimate Proportion Invariant; HKY model; Estimate Kappa.

b. Clock Model Tab: Select Strict Clock; Clock.rate: 0.01035 (This is equivalent to 2.07% pairwise divergence following Weir & Schluter Mol Ecol. 2008. 2.07/2/100=0.01035 per lineage rate)

c. Priors Tab: Set Birth Death Model as tree prior. Leave other prior options with the default values.

d. MCMC tab: Set chain length: 5000000 (i.e. 5 million. If it’s too slow on your computer, you could reduce this to 1 million - the longer you run the MCMC for the most likely it is to converge); Tracelog file name: Cyanistes.log; Tree log file: “Cyanistes.trees”

5.1.3 Save the XML file.

In File: Save as file “Cyanistes.xml”

5.1.4 Run XML file in BEAST.

Open BEAST 2 and load “Cyanistes.xml” file that was created in BEAUti 2. This will run the MCMC chain – this will take several minutes.

Note: Normally you should run multiple independent chains and combine the results from several runs using software “LogCombiner”, but for practical purposes we will run a single chain.

5.1.5 Check for convergence of MCMC run.

Once the BEAST run is completed load the Beast output ‘Cyanistes.log’ file to software Tracer. Here you can check whether the diagnostics of the MCMC run are good - if ESS values are low (flagged in red), this means your run did not converge; and the under the “Tracer” tab, the plot should look like a “caterpillar”, without abrupt changes. For the purposes of the practical, you can proceed even if the run diagnostics are not great (would have to run a longer chain).

5.1.6 Generate maximum clade credibility tree.

Load Beast output ‘Cyanistes.trees’ file to TreeAnnotator (included in BEAST software folder) to summarize the results of the posterior distribution of trees into a single tree (maximum clade credibility tree)

Select:

a. Target tree type: ‘Maximum clade credibility’;

b. Burn-in percentage: 20%

c. Node heights: ’Mean heights”

d. Output file: “Cyanistes.tre” - Make sure you select the folder you want to put it in by choosing “Choose File” and select the folder you want to put it in, otherwise you might not be able to find it! Best is to write it to the same folder as the input file.