Overall, I think physics is lucky to have its premier journals (Physical Review) be run by our own nonprofit professional society—APS. I think that explains, at least in part, why the arXiv has been so successful in physics and why similar efforts have floundered in other fields.
All that said, I have one bone to pick with the Physical Review journals: they insist that footnotes should be denoted in the same manner as citations [1,2]. Citations and footnotes serve very different purposes, and I both use and consume them in very different ways. When I’m reading a paper, I often read the footnotes, especially if I’m trying to totally understand a passage. I almost never look at the citations on my first reading. As a reader, I love footnotes! They’re a great way to add context, clarifications, parenthetical remarks, or definitions without interrupting the flow of your argument. Citations, on the other hand, are for backing up your claims or giving proper credit. If you mistake a footnote for a citation, you might miss some useful information or you might wrongly assume that the claim is backed up elsewhere in the literature [3]. Finally, I prefer footnotes to endnotes because footnotes keeps the information nearby, rather than forcing readers to flip pages back and forth.
In summary, Physical Review Letters and A, B, C and D: please follow Physical Review E‘s lead and allow separate footnotes!
Just one section called references:
[1] Waldron et al. “The Physical Review Style and Notation Guide” APS 2011 (a citation)
[2] At least for Physical Review B and Physical Review Letters, which are the PR journals I use. Phys Rev E does allow separate footnotes. (a footnote)
[3] For example, if you mistook [2] for a citation, you might not have noticed my caveat about which specific journals exhibit this problem.
Update 2020-03-12
Let the record reflect that after I posted this I heard that PRB does, in fact, allow separate footnotes. I tested this with my most recent paper and I now have experimental proof.