At first glance, the difficulty I see with the information at the Byrd DNA Project is that the very basic, user-reported lineages aren't matched to the individual test kits. For example, there are four kits in "Haplogroup G - Lineage 1," only two of which name a believed earliest known ancestor. The pedigrees at https://www.familytreedna.com/groups/byrd/about/results don't offer the ability to determine which test kit is associated with which pedigree; they're all simply clustered under the Group heading.
An aside is that the text we see there in bold red (e.g., B-26, B-49, SM-1) is almost certainly from the two-years-defunct WorldFamilies.net. I suspect that's where the pedigrees came from, and those red-text designations used to be what associated test kits with pedigrees.
Without knowing which kit claims to match which lineage, we can't really make any progress there. And FTDNA offers contact capability only to customers who match each other, and to the Project administrators, providing the test-taker has granted them that level of access. It's going to take some digging. Requesting via one of the Project admins that your email address be given to the relevant test takers in order to compare notes could be a start.
But there's an additional difficulty even within the two G haplogroups Lineage 1 and Lineage 2. The problem is that the Project admin has, correctly based upon the STR (simple tandem repeat) values, broken them into two distinct groups. By my cursory look, and using the modal values for "Haplogroup G - Lineage 1," at 37 markers kit 101394 in Group 2 would be a genetic distance of 11, and kit 312542 would be a GD of 13 from the members in Lineage 1. Family Tree won't call 37-marker results as a match at all if they are more distant than a GD of 4.
While nothing is impossible, based on those STR differences I would presume it extremely unlikely that the two sets of men shared a patrilineal ancestor in the genealogical timeframe. Their common ancestor would be several hundreds if not more than a thousand years prior to the ancestor dates shown. We really can't make anything of the SNP data unless at least one of the test-takers in "Haplogroup G - Lineage 2" takes a Big Y test. Haplogroup G currently contains 1,262 branches, and 1,124 of them are under G-P15, the deepest tested SNP in that Lineage 2; so kit 101394 being P15 positive doesn't tell us much.
"Haplogroup R1b-Lineage 5" looks to be internally consistent, again at a quick glance only, but none of the men have done deep testing with the Big Y.
Given the above, I think there are at least three genetic disconnects in these lines, not two. Each of the three groupings as the administrators show them on the detailed results page represents a distinct patrilineal line. So I think I may have managed to make it more complicated, not less.