The frequency in isolated populations of primarily non-Slavic communities strongly suggests that it would have had to have been introduced in pre-Slavic times. Armenian levels of I2a and Greek I2a1* suggests that I2a may have been part of an early Illyro/Phrygian exchange, Phrygian being the Armenian I2a so prominent today.
The King paper showed significantly higher levels of I2a in the Peloponnese compared to North Greece. The King paper found I2a2a (Former I2b1 in the Y2010 tree) at relatively low frequencies throughout Greece from the North to the isolated Lasithi Plateau on Crete. The Battaglia study found a sturdy 17.5% of Greek Macedonian men to be I2a1*.
(1) One should not confuse I2a-Din with I2a
(2) The Slavs absolutely swamped Greece in the late 6th and 7th cs. From top to bottom. The non-Slav population was pushed back to meager coastal areas and a few large cities. The Byzantine reconquista which began later in the 7th c. was largely successful as we know (due to Slavic disorganization), but their earlier massive presence explains the current level of I2a1b1 in Greece with no difficulty at all.
(3) Ditto re the large amounts of I2a-Din among Romanians, some Albanians, Vlachs, and Moldavians. All the result of the Slavic deluge. These early Slavs were not impressive state builders, but Jordanes already noted their impressive numbers. They settled in the Balkans both compactly and amongst other populations. The fate of these myriad little communities was subsequently determined by internal processes. Slavs assimilated non-Slavs linguistically, or the reverse. But the genes remained.