Thanks for your quick response. I was linking some PDFs that I already had on my hard drive and so wasn't downloading from the internet into Zotero... but downloading a new PDF from Springer using the Zotero Chrome Extension in the browser seemed to work as you suggested. Thanks :-)

From my personal experience this is sometimes a Zotero problem (when Zotero detects the wrong DOI on the webpage). However, it is frequently a Cross-Ref / publisher issue. I use a clunky in-house utility to help me recognize records that have missing volume, issue, page range metadata. It depends on the DOI to reach the publisher and tests for updated metadata. We had to build in title and author tests else the missing metadata would be replaced with that of some other article. Yes, this can happen with Zotero downloaded metadata; but it also happens with direct ftp feeds from publishers. Even the DOIs that come directly from publishers sometimes fail to connect with the correct article.

This problem can be compounded when a journal's publisher changes. It isn't clear to me exactly what is supposed to happen but the old DOI to an article in a publisher's back files can link to the original article or not -- leading nowhere. Often, the articles receive a new DOI. For some reason articles -- even when there has been no change in publisher continuity -- can have more than one DOI assigned. This multi-DOI policy led to my great disappointment when my idea of using DOIs as an absolute test for duplicate records proved to be less reliable than I hoped.

edit

There have been times that we needed to change all of the old publisher's DOIs to the new publisher's DOIs. I don't know if the problem is with the old or new publisher but it sadly defeats the idea of durable linking.