Distinguish three ways in which we might end up with multiple senses of "ought":
(1) Most obviously, perhaps, are the content restrictions or "end-relative oughts": prudential oughts, moral oughts, etc. But these are of limited normative significance -- to think that there are only end-relative oughts would amount to nihilism by another name. So I prefer to focus on what, if anything, we really ought to do ("all things considered", as it's sometimes said). If there are real, categorical 'oughts' of this kind, then they clearly take priority over the various more restricted or end-relative 'oughts', in which case there's no deep pluralism to be found here.
(2) Evidence-relative oughts. Here I think there are two genuinely normative concepts on offer: the "objective ought", or what would be best in light of the actual facts, and the "rational ought", or what is recommended in light of the agent's available evidence. Some might think there are even more "subjective" variations, all on a normative par: what the agent ought to do in light of their non-normative ignorance only; what they ought to do relative to their false (non-normative and) normative beliefs; what they ought to do relative to their false (non-normative and) normative and logical beliefs, and perhaps various other mental incapacities, etc. But I'm inclined to think that just one of these "subjective" oughts is privileged by its proper action-guiding role and connection to praise- and blame-worthiness. Whichever it is, that's the one I mean by the "rational ought". (I lean towards counting only non-normative ignorance/uncertainty, but it's a substantive question whether normative and logical ignorance/uncertainty are also rationally relevant.)
(3) As suggested in my satisficing paper, we might want different "oughts" to represent different levels of demandingness. Again, I'm hopeful that we can restrict this to just two: the fully-ideal "ought of most reason" vs. the more permissive "ought" of minimal decency (or non-blameworthiness). But if my argument in that paper fails, then I could see a Scalar Consequentialist thinking that one could stipulatively introduce any number of more-or-less demanding 'oughts', none of which would be privileged over the others.
So, in the end I'm inclined to think there are four kinds of 'oughts' worth thinking about: all categorical and unrestricted in content, they might be either 'objective' or 'rational', and each of these might concern either the 'ought of most reason' or the 'ought of minimal decency'.
If I wanted to be a monist, I'd probably favour the categorical, rational 'ought' of most reason. As I argue in my paper, the more permissive 'ought' of minimal decency is of secondary significance -- so I guess one who doesn't mind a stark departure from common moral practices could construct a perfectly sensible theory without it (e.g., traditional maximizing act consequentialism). And the objective 'ought' could presumably be replaced by the evaluative concept of what's "best", or perhaps reduced to what would be rational given (momentary) full information (modulo worries about the conditional fallacy).
Do you agree with this break-down? Are there other potential sources of "deontic pluralism" that I've missed?