The paper states this is a new sample from Kotias Klde that's J2b. At this point, it's only a suspicion that he may be J2a, as the dating, site, mtDNA, is exactly the same as previously published KK1 which is
J2a-Y12379. So it's possible he is some paternal relative of the J2a sample and the software got a false positive for a J2b SNP or something like that. The coordinates are reported a few miles apart but they should be in the same general site, as Kotias Klde cave doesn't cover an area spanning a few miles.
We've emailed the authors to see if they can provide further clarification, but so far no response. The reported coverage at 3.6x is very good for an ancient sample, even better than NEO806 (the "early Messapian"). So we will know for sure what he is once the raw data are published, along with the deeper classification. I'm personally thinking this is a new J2b sample, perhaps even an "early" M241>L283, but we'll see..