Here's the gist of Prothero's criticism:
Yet FPP take the Gould/Lewontin critique too far, and make the absurd claim that because some features are possibly constrained and not fine-tuned by natural selection, we cannot assume that natural selection works anywhere. What about all those studies that demonstrate tight correlations of cause and effect between a feature and the selective response that occurs when nature intervenes? According to FPP, these are not conclusive enough. Therefore, whenever we have a large data set that shows a strong correlation between say, obesity and heart disease, or increases in carbon dioxide and global warming, we cannot even begin to suggest that there might be a causal connection. If this is the angle that FPP are pushing, then they have a dispute with almost all of science, not just evolutionary biology. [Emphasis added.]I can only second that. The inference that we cannot assume that NS works anywhere is absurd, because i) it is very, VERY well understood theoretically, and ii) as Prothero points out, there are many studies that show "that have done careful work with controls and minimizing the variables that conclusively show natural selection to be the only reasonable explanation for the results".
* More on Donald Prothero and his great book, Evolution: What the Fossils Say and Why it Matters, here, and on debating creationists here.