Innarestin', CJ, thanks.
Though the hypothesis concerning size evolution on islands was hardly "magical" thinking, since there was a clear naturalistic rationale given for the claimed effects (predator-free smaller animals no longer being selected for their ability to be unobtrusive; larger animals being selected against due to limited resources).
Also, the abstract for the article
http://www3.interscience.wiley.com/journal/123356145/abstract?CRETRY=1&SRETRY=0specifically states that the researchers did
not rule out island dwarfing (as suggested for
H. floresiensis, for example) -- in fact, to the contrary:
We did find consistent evidence that large (> 10 kg) mammals grow smaller on islands. Smaller species, however, show no consistent tendency to either dwarf or grow larger on islands.
My bold.
To this extent, at least, the article's conclusions are over-extended in the
Science Daily discussion.
And the way the *magical* island-size hypothesis that they were testing was expressed is a much different formulation than the typical way of expressing the smaller-get-bigger-bigger-get-smaller generalization that I have encountered:
Optimal body size theories predict that large clades have a single, optimal, body size that serves as an evolutionary attractor, with the full body size spectrum of a clade resulting from interspecific competition. Because interspecific competition is believed to be reduced on islands, such theories predict that insular animals should be closer to the optimal size than mainland animals. We test the resulting prediction that insular clade members should therefore have narrower body size ranges than their mainland relatives.
I have frankly never seen the claim that there is some environment-free "optimal" size for any given clade, and certainly the claim that the environmental constraints keeping some animals small are loosened on islands is not easily transformed into a claim that there is an "optimal" size that the constraints are suppressing.
Since the article itself is beyond a paywall for me, I don't doubt -- or, at least, have no basis for refuting -- that the authors supply cites to support their characterization of the "optimal clade size" hypothesis. I've just never heard of it being expressed that way.