The region may affect the concentration range need, but the size of the amplicon is usually more important. Since primers targeting different regions generally amplify different sized amplicons, it will appear as though the region matters more than it does.  The community you are studying can also affect the gradient range needed. 

Usually what I would do is run a range-finding gel first, with a few samples on a gel with a denaturing concentration from 20 to 80%.  When you visualize the gel, you will probably see that all the bands are found in a portion of the concentration range.  For the rest of your gels you can use the narrower range suggested by your range-finding gel.

I would not use the primers mentioned in your third link, as they are meant for sequencing, not DGGE.  The amplicon is 1.5 kB, while DGGE primers will usually amplify an area of only a few hundred bases.  Also, DGGE primers will have a GC clamp on the forward primer.  It would be better to use the primers mentioned in the second paper as they are meant for DGGE.

Thank you for that answer, it clarified things for me and I'm embarrassed that I did not see that the 3rd link's primers were for sequencing.  

Selecting the variable region will affect you output. So you must decide if you stay with V1-V3 or V6-V8 region; you can also try both. If you don't know the range, you perform first run with wider range (as Alison suggested) then you can narrow it. 30-60% works in most cases pretty well; however if you expect shift towards Bifidos, extend it up to 70-80%.

If you're intersted in a particular group you can also perform sandwich PCR and then run the DGGE for e.g. Lactobacilli or Bifidobacteria.