evolution - Why does the gc content deviate from 50% in prokaryotes

I have read quite some articles but I can't figure out the main reason for gc content deviation in prokaryotes. In eukaryotes I can understand it, because the genome isn't composed at random, like TATA boxes and CpG island, because they are important for functioning in the production of proteins.

However in prokaryotes there is such a variety, GC% is ranging from 20% up to 70%. Most of the time it depends on the environment, high temperature needs a stable genome (= high gc content).

I also read the answer to this question How does GC-content evolve? however it's still not clear to me. I hope someone can explain the deviation a little bit more.

Question

Prokaryotes have an AT drift, but what mechanism causes some of these bacteria to higher their GC% instead of lowering it.

Answer

I wonder whether it is just some sort of functionless drift. I don't know about bacteria, but think the example of some human viruses may be instructive. The human herpes simplex 1 virus (causes cold sores) has a very high GC content, whereas the quite closely related (in terms of gene repertoire and organization) human varicella zoster virus (causes shingles) has a bias towards AT. Now these viruses depend completely on the host cell's translational machinery, which will be similar (e.g. distribution of tRNA species) in the cells these two viruses infect (as will temperature, ionic strength etc.) (The viruses encode their own DNA replication enzymes but not the systems for transcription or translation.)

So if there is a functional reason for the differences in these alpha-herpes viruses, it is difficult to discern, and one might assume that the situation in bacteria might well be similar.

Clearly, there must be mechanisms for the accentuation in a GC bias once established. One imagines there could be a selective advantage in either replication, transcription or translation, but molecular explanations do not jump to mind (at least not to mine).