Neither?
https://www.genealogy.com/forum/surnames/topics/landrum/2599/
“I cannot answer your question completely, but I know for a fact that the Mary Chew who was a DAUGHTER of Larkin Chew, Junior and his wife Mary Beverly married John Smith (a son of Maurice and Elizabeth Smith of Middlesex Co., VA and then King and Queen Co., VA....this was a branch of my mother's Smith family).Also, records seem to clearly show that the SISTER of Larkin Chew, Junior, who was also named Mary Chew, married Robert Johnston.So....unless there was yet another/third "available" Mary Chew in the correct age range in this overall Chew family, it would appear that John Landrum did NOT marry a Mary Chew.Like you, I have also seen the wife of John Landrum listed as Mary Buckner, but I have done no true independent research on that whatsoever. ...”
Well, I'm an advocate of two things:
1) Parents who can't be documented shouldn't be connected. For example, I was tracing the line of a distant cousin when I hit this bump. I had to stop here because i can't trust that the rest of the pathway is even related. I'd rather see no parents than 'ify' or actually incorrect parents connected that give false pathways.
2) There are 15 profile managers attached to this profile. Surely some are direct descendants or at least related and should be the ones to do the research and sort this out and fix any errors. I have enough of my own lines to do that on.
That's just my opinion on such issues. I just point them out when I run across them in hopes the profile managers will care enough to participate in making sure the lines are correct so that the next time I'm breezing through on a cousin line I don't have to stop in my tracks for the reasons given.
Add to that the link you sent, and I'd say a disconnect is appropriate. Then if the relatives want to rebuild the line correctly, that's awesome.