constrained to be monophyletic (for a protocol to carry out
the simulations, see Van der Niet et al., 2005). These data sets
were subsequently analyzed using parsimony implemented in
PAUP⁎4.0b10 (Swofford, 2003) with 50 random addition
sequence replicates, holding three trees each step and saving no
more than five trees. The difference in steps between a parsimony
analysis constraining Satyriinae to bemonophyletic (i.e. the same
topology that was used to simulate the data) and the most
parsimonious tree for each of these 100 data sets was recorded.
The observed value was contrasted to this null distribution
generated through simulation. If the observed value does not fall
out of the 5% tail of the null distribution, the hypothesis that a
difference in steps between a tree with a monophyletic Satyriinae
and the most parsimonious tree is caused by the stochastic nature
of the substitution process cannot be rejected.