As explained in the above paper, the PS-I complex is primarily located within non-appressed thylakoid region and stroma lamella, while PS-II in the appressed regions. Now since non-cyclic electron transfer (non-cyclic photophosphorylation) requires both the photosystems, wouldn't their distribution within grana far from one another hamper their ability to carry out non-cyclic electron transport. Although there are mobile untis to account for the electron transport between marginal PS-I and the PS-II immediately below, but the electron transport to the PS-II units deep in the granal core would still be unaccounted for.
First a clarification: the mobile units are no so "mobile". Due to (1) the high protein:lipid ratio of the thylakoid membrane and (2) the packing of protein complexes, the rapid diffusion of components along the membrane is hindered.
It has been proposed the existence of small local PQ diffusion microdomains, based upon the percolation theory:
For this reason the PQ are confined, on a short time scale, within these domains, preventing their "dispersion".
In addition of the low diffusion of mobile carriers, the geometry of the granal stacks mantains the contact area between the PSII of the appressed grana and the PSI in the margin annulus domains relatively large, enabling effective linear electron transport in microdomains.
Figure 2b of the paper cited below shows the geometrical organization of the stacked grana and its description says:
Further informations can be found in Albertsson, P-Å. (2001) Trends Plant Sci. 6, 349–354.