In order to move this debate to a level that considers the methodological problems as well as the philosophical ones, since Wagner and Marshall both espouse the stratolikelihood approach, I think that the method deserves fuller exposition than either has provided. Moreover, some sober second thought could be in order as well.
Consider a set of stratigraphic ranges labelled here arbitrarily 1 through 7. If we assume that each fossil species represents a branch of the tree (terminal or internal branch), then we can variously arrange the ranges and see which one minimizes the amount of gap (red lines) as is done here for trees 1, 2 and 3.
Depending on the sampling density for each lineage, one or another topology will be considered more likely. Assuming equal sampling densities, tree #2 is preferred of those above.
So far so good, but what is not considered in the method as Huelsenbeck and Rannala1 devised it is that, if you are going to allow ancestors, one lineage may be the direct ancestor of another but is separated from it by a gap due to any of the reasons detailed by Holland2.
In tree #4 to the right, 5 is ancestral to 7, and 4 to 6 though there is no branching between these ancestors and their descendants.
I see no reason why tree #4 is not tenable while tree #2 is, however, tree #4 is not one considered by the stratolikelihood method.
"Species" 2 and 3 each are paraphyletic. That is, for example, 2 is ancestral to 5 and itself, and then later is ancestral to 4 and itself and also is ancestral to 7.
The point here is that stratolikelihood does not consider this possiblity even though it is tenable and more optimal than any other arrangement.
I am having trouble understanding quite why Wagner and Marshall
are so much in favour of a method that is so clearly misleading.
Mark E. Siddall
References
In fact, because we are allowing for ancestors, and admitting that minimizing the amount of gap (red lines) is our optimality criterion, then tree #5 to the left does a better job than any of those above. However tree #5 has an interesting problem.
Museum of Zoology,University of Michigan,USA
