I suspect it may simply come down to the size of the photosites on a crop sensor. They tend to be smaller than those on full-frame sensor and thus the laws of physics and diffraction come in to play sooner on a crop sensor.
For a fuller and more in-depth answer just hang on, I see our friend Neuro on the horizon with his usual top-notch explanations for the witchcraft that is optical physics.
Pah! Just as I type this, there he comes!
I have a question also. Diffraction. I have read about it, and there are several places that warns against closing the lens too much (some say you should not go to F22, but stay at F16 due to diffraction. Does anyone have a practical example on how this would alter a picture, besides a theoretical argument? Otherwise, when shooting landscape for instance, why would one not go minimum aperture all the time, given that you control the other factors?
Whilst it was only speculation on my part, check out the examples here:
Also the table near the top of the page here http://www.the-digital-picture.com/Reviews/Canon-EOS-7D-Digital-SLR-Camera-Review.aspx gives values for apertures where diffraction might start to creep in. Note that they say it may not be visible depending on what size and how you view the image.