The illustration being:
http://eligius.st/~artefact2/blocks/1PvLowa5f59HXfeEyTrBAog6ybosdn4qMEUp until around 4 days ago (approximately block 131092, right before the last US block), the number of invalid blocks seems to be somewhat in line with deepbit's estimate of 1-ish%, then look at what has happened since.
It would seem extremely unlikely that the US pool going haywire at the same time the number of invalid blocks has increased around 20-fold are unrelated.