I read the review in Science with much appreciation as well. Have you looked into much about likelihood for testing hypotheses? I think that this is extremely relevant, especially if you view your test as trying to reject hypotheses from a set of “candidates” and giving weight to the others. You don’t need a p-value when you are just saying that one possible hypothesis is better than another possible (not null) hypothesis.

Tom Hobbs, a professor at CSU, has been pushing this view in his work (on ungulate behavior) and he teaches it in one of his graduate classes. He has a graph where he suggests the growing emergence of new statistical techniques (especially Bayesian statistics) in the major ecology journals.

Like you said, any cultural change is hard. I think that ecologists and other scientists are starting to catch on, though. And it certainly helps if leading statisticians are on the same wavelength (that you shouldn’t need p-values to show your point).