Relationships, or not

Continuing our discussion from yesterday concerning what we might discover from the Project-wide genetic distance table comparison   -   that area comprising the Faires Berrys, Rockingham Co. Berrys, Barry/Berrys and New Jersey Berrys has always seemed almost interrelated, as illustrated here -

For the purposes of this discussion,however, I think it might clarify matters somewhat if we remove those two 12 marker haplotypes since they tend to indicate more false positives.
Well, that fairly clearly separates out the Barry/Berrys and New Jersey Berrys but still leaves the Rockingham Co. Berrys and Faires Berrys groups somewhat unsettled, probably primarily because of tbd who indicates clearly to be unrelated to Robert at a genetic distance of 9 but related to Kevin with a genetic distance of 3, Robert and Kevin being related Rockingham Co. Berrys at a genetic distance of 2 plus each having a paper trail to a common ancestor.

Anyone have a view on what might clarify these relationships?    I don't have the feeling that simply adding more markers would necessarily do it although upgrades to 37 markers by Dennis, Paul and Edward might.    They do seem to be the source of all those 'probably' blocks.    There may be some strategically selectable markers that would help but we don't have enough data to know which those might be.