Analysis things to do

[willpearse]

• Do a quantile regression on figure 1 (the infamous focal ED correlations). Check the lower, upper, and middle quantiles (25, 50, 75? ish?) to show that the value is 'converging' on 0 (ish)
• Drop number of species in focal clade down to 4 or 5; justify in methods because we're calculating correlations. I don't think we need to do something special for clades of size two (...I could be wrong...)
• Increase number of species in dropping simulation to 75% (why not?)