The methodology is explained in
Achaz G, Palmer S, Kearney M, Maldarelli F, Mellors JW, Coffin JM, Wakeley J. (2004) Mol Biol Evol. 21(10):1902-12
The syntax of your sequence names must take into account that there are subpopulations in the set.
In order to discreminate automatically the different groups If not a chronological serie, the order in which the populations appear in the fasta alignment is important for the graphic representation. The different populations names are taken in order of seen in the alignment and are spotted in the same order on the graphic. You can change this order if needed at the end of the form, you have to add at this end of each sequence name, separated by a semicolon, the name of the population.
e.g.: for a sequence belonging to pop 1
if your data are chronological, choose a int value at end of pop name and separate it from the name by a dot (ex here day=136)
For now, sites with indels are completely ignored.