Now that you mention this issue, it wouldn't surprise me if all of the samples are from a single site, so while I think the Yamnaya population as a whole was a mixture of R1a, J2, G and possibly R1b, those few samples may not be representative and may skew toward one or two haplotypes, depending on how well mixed the Yamnaya population was. And the bias toward R1a in northeastern Europe is either mostly Corded Ware without as much IE impact as we assumed or a result of the more northerly Yamnaya being biased toward R1a because they were less mixed with the "Armenian" component. So, depending on where the dividing line was for the Yamnaya population, these few samples could be all R1a or no R1a, or a mix.