Mark's statement about high levels of P and K being "locked away" in the soil matrix is true, and has been touted by a large number of companies trying to sell brand-name fertilizer or special potions to "stimulate" your soils and release this source of plant nutrition.  -- The only problem is that it doesn't work that way.  I know of several people who went down that road, and eventually they mined their soils of P to the extent that many crops grew poorly and plant tissue analyses showed low P to be the culprit.  Scattering a bit of P fertilizer and observing the results confirmed this.  They now must bear the pain of having payed 2X or 3X per unit of P for those special sources, have lost yield over many years, and now must pay again to b'cast some 11-52-0 to get it straightened out. 

Maybe it works somewhere, but it sure as heck doesn't work out here on the Plains. 

As for GTD's post, I would concur on the general points although I don't know where the percentages came from (perhaps just illustrative in nature).  The value of band-applied fertilizer (near a crop row) is insurance that it will have enough, especially early in the season.  Occasionally, crops show some response even on high-testing soils.  However, you cannot rely long-term on cutting rates by banding.  The nutrients must come from somewhere (see first paragraph).

Removal per bushel does vary depending on whether the plant was under-nourished or not, as well as by hybrid or variety, and the growing season.  But the efficiency claims by banding (near the crop row) has to do with amount applied that year in relation to yield, and wouldn't have much effect on quantity of P or K per bushel within the normal range of nutrients supplied and crop yields.  If anything, the banding might cause higher removal per bushel due to improved uptake.