I am curious: what steps do generalist journals follow to try to ensure that every subfield (e.g., number theory, differential geometry, combinatorics, etc. for pure mathematics) is adequately represented, as the journal deems suitable?
To ask the question in a different way: a generalist journal will typically publish many more papers in one subfield compared to another subfield (in an absolute sense and/or relative to total volume of papers in that field). How does this "publication rate by subfield" come to be? My personal interest is in mathematics, though the question applies to any academic field.
There are two sub-questions here: one for high-volume, "good" journals (e.g. Proceedings or Transactions of the American Mathematical Society) and one for low-volume, "top" journals (e.g. Annals of Mathematics, Acta Mathematica). It is a given that such a journal would have a diverse editorial board (in some cases, subject to constraints such as university affiliation), which is already a partial answer to the question. So the question is whether journals conscientiously or systematically do more than this.
My understanding is that for a journal like Proceedings or Transactions, each subject editor has a certain page quota of articles that they can recommend for publication. What I don't know is whether such a page quota is the same for each editor, or whether this varies by subfield. Or how they decide how many editors per subject. Presumably one subfield may be significantly larger than another, or a single editor might end up being the contact point for various subfields. I also don't know how final decisions are made based on the recommendations of subject editors.
For the very top journals, the discretion of the editors would presumably be a more important factor. But is "we've already published in this field recently" or "we haven't published in this field recently" often a consideration?