Hi Mary, I have only been following this from afar, and may have missed some context. Just to say that, from the history of the BasisOfRecord term, the main intention of the term has been to allow a user of data to distinguish between in-situ (specimen, observation) and ex-situ data (botanical/zoological garden, culture collection), as well as between vouchered (specimen) and unvouchered (observation) data. Fossils are in because knowing this is required to interpret spatial/temporal data attached to them. MaterialCitation was a later addition to mark literature-based datasets (for a while, we had a term called “Literature”). When in doubt, I would recommend to stick with the DarwinCore documentation (here: Darwin Core List of Terms - Darwin Core).
You are correct that there will be derivation chains (an observation captured in a field book, later aggregated in a local checklist, which is then published in a book, that someone later extracts distribution data from for a database, etc). When deciding which BasisOfRecord to give a record, I find it helps not to overthink this history, and rather be guided by what users would like to be able to filter for: here, that data was transferred from a literature source / printed publication, and thus several interpretation steps removed from the original occurrence. If specimens exist, they would ideally have their own database representation, and linking between the MaterialCitation record and the PreservedSpecimen record would be perfect. Without that, it is good enough for the user to know that the data they are interpreting stem from a literature source.
I am not sure whether this helps, please let me know if you need more