Context: I do volunteer work doing data entry for an online database of organisms name (not Wikispecies, though I have edited there too). When entering data from published catalogue catalogues, it is not uncommon for some issues to be unearthed. Usually I'm happy to signal them to the authors, even if the corrections never actually see print.
The paper I'm currently handling, however, seems to have an unusually high ratio of errors. I have handled two pages of names out of ca. forty and already I have found that seven references were missing entirely from the reference section. Combined with various other issues, I feel like this warrants more than merely contacting the authors with my findings.
Additionally, I know the authors have access to a database that lacks the errors in question: I've used it to find the problematic references to begin with! So signaling the errors is not likely to be seen as useful to them. However, I'd like to avoid other readers falling afoul of these problems I have to solve, so it's kind of pointless if it never sees print.
I'm not sure I'm confident with either contacting the authors or addressing a corrigenda (which I expect to take a while to compile anyway, I have to handle of 1.4k names here) directly to the editor. Am I overlooking some third options or should I just bite the bullet?