Supported constraints agree more with the molecular clock than unsupported constraints

The three following plots show that the most highly supported constraints selected by MaxTiC tend to agree more with sequence divergence as measured by d[donor]-d[recipient], where d=clade-to-tip divergence.

I think similar plots could be used as an alternative to the new proposed figure 2, as a way to convey the following points:

• individual constraints tend to agree with sequence divergence, in particular supported constraints

• some outlier constraints exist: those constraints that are highly supported yet disagree with sequence divergence are potentially interesting in that they may reveal areas of the trees where the molecular clock is mistaken.

Such a reasoning could be used as a transition to the discussion of the biologically relevant constraints.

Useful functions

Cyanobacteria

## [1] "Total number of constraints:  699  ; MaxTiC constraints:  592  ; Rejected constraints:  107"

Boxplots by quartile

Archaea

## [1] "Total number of constraints:  1889  ; MaxTiC constraints:  1477  ; Rejected constraints:  412"

Boxplots by quartile

Fungi

## [1] "Total number of constraints:  1980  ; MaxTiC constraints:  1479  ; Rejected constraints:  501"

Boxplots by quartile