The problem is, if you don't have a FR curve to reference to after each adjustment, PEQ or Graphic, it will be a hit and miss thing to some extents. For example, you can select a +3 gain to 70 Hz with a Q of 0.5, the overall result may be worse than if you don't, and the dip at 70 Hz observed before may not change much at all after the adjustment. I did play with PEQ using REW's generated equalizers for a miniDSP 4XHD and that should be better than just manually adjusting them on the Yamaha. So what I am saying below is in based on my user experience, that seems to be consistent with what I read about how various REQ filters work.
Again, I don't know enough to explain the "why" part, but I do know if you do it manually by adjusting the gain and Q using Yamaha's interface, the resulting curve will most likely not follow the target curve too closely. Then you will have to keep playing with the gain and Q, try going in both directions, and then run the sweeps again to watch for improvements. You may have to do it in incremental ways many times, and may not get too close to the target curve, but hopefully closer to it. Using the Audyssey Editor App has similar issues. For example, if you shape the curve in such a way it ramps up 2 dB/ octave (just an example) from 100 Hz towards 15 Hz, the resulting curve may follow the trend generally, but it won't necessarily be 2 dB/octave. So it helps a lot in general but don't expect the actual results follow the target exactly.
Audyssey (FIR type filters, Dirac Live (mixed FIR and IIR type) should, at least in theory, be more effective as they are inherently more powerful. They do need powerful DAPs to run well, and that may be the reason why XT32 is only offered in the higher D&M models, that have more powerful DSPs.
I think you are wise to apply PEQ to the 30-80 Hz range because that obviously would simply the process and take a lot less time to achieve satisfactory results than to go all the way to 200 Hz or higher.