There are at least three possible explanations (and quite likely more):
1) You may be correct in suggesting that the OTT isn't "very coherent."* In other words, they may have a little work to do. I don't know enough about the site to offer a solid opinion.
2) The OTT may be unintentionally repeating someone else's error(s).
3) The OTT has to work with the information that's available, and, in the realm of taxonomy, that information can be very confusing. Different authorities recognized different classification schemes. Scientific names also change over time; for example, the polar bear (Ursus maritimus) used to be Thalarctos maritimus.
I don't know enough about OTT to really understand exactly how it works. However, I typed in Ursus (a subspecies in the bear family), and it listed 22 "Descendant tips." Curiously, it didn't display a list of species and subspecies, however. When I type in Panthera (the "big cats"), it does display a list of species and subspecies. I don't know if that reflects a problem with OTT or not.
You're focusing on the genus Rosa, which, as you say, is complex. In fact, I just thought of a fourth possibility...
4) Domestication - Many species in the genus Rosa have been domesticated and/or hybridized. It's possible that OTT is including the names of cultivated species. (It shouldn't be too hard to determine if this is the case if you're familiar with the Latin names of some cultivated species.)
Ultimately, the people best qualified to answer your question are probably the people behind OTT. I'm very interested in finding a definitive answer myself, because taxonomy is a major personal interest. I also find it endlessly confusing. ;)
P.S. I once wrote an article about a particular phylum of marine invertebrates. I stated that there were X number of species in the phylum, using a similar website as a reference. However, a specialist later told me I was way off base. I can't remember the details, but I believe there was either an error with the "master database" or there were conflicts between two or more databases. Like I said, taxonomy is confusing.
*Regarding coherence, we might also consider the target audience. This definitely is not what I would call a user-friendly site. I'm a lifelong naturalist with a degree in ecology/biology, and I find it very confusing. But rather than call it "incoherent," we might just say it isn't user-friendly.