I think it is time to recapitulate the facts:
- The highest I2a-Din frequency has been measured by Battaglia et al. (2008) for "Bosnian Croats" at 73.3 %. While the text does not give details on where the samples were taken, the enclosed map suggests Mostar as sampling region.
- The next highest frequency is reported for the island of Hvar by Barac et al. (2003) at 65.9 %.
- Pericic et al. (2005) report Herzegovina at 63.8 %. This relates to combined sampling from Mostar and Siroki Brijeg. On the assumption that the frequency in Mostar is around 70%, as suggested by the Battaglia result, we can infer a frequency around 55% in Siroki Brijeg. Frequencies above 50% have furthermore been reported from Zenica in central Bosnia (52.2%, Pericic et al.), and the islands of Brac (55.1 %) and Korcula (53.7 %).
- Next comes "Bosniacs", reported by Battaglia et al. at 45.3% - the enclosed map suggests Sarajevo as sampling location. Western Montenegro may have similar frequencies, as is discussed here http://www.eupedia.com/forum/threads/26949-montenegro-dna.
- Frequencies around 30% have been reported for Bosnian Serbs (34.6 %, Battaglia, the enclosed map suggests sampling around Tuzla), mainland Croatia (Battaglia 33.7%, Pericic after deeper genotyping of Barac's data 32.2%), Belgrade (29.2%, Pericic), Skopje (29.1%, Pericic), Krk (28.4%, Barac), and Ossijek Croats (27.6%, Battaglia). For the Croatian mainland, Barac mentions a higher I frequency in the Southern and Eastern parts.
That is a strikingly regular, almost concentric pattern with its "epicentre" close to Mostar, and frequencies above or close to 50% in an area that is approximately demarked by the cities of Herceg Novi, Sarajevo, Zenica and Split.
Now, let's take a look how the high I2a-Din distribution has been affecting other relevant Y DNA haplogroups:
- E-M78 has obviously been largely unaffected by the I2a Din concentration. Generally regarded as Albanian marker on the Balkans, the E-M78 frequency peaks at 45.6% among Kosovo Albanians (Perisic) and 36% among FYROM-Albanians (Battaglia). (South-)Westward from there, it declines steadily from a 27% average (with a westwards decline) in Montenegro, 20.4% in Belgrade (Pericic), 19.8% in Tuzla (Battaglia), 13.1% in Sarajevo (Battaglia), 10.1% in Zenica (Pericic), some 9% around Mostar (Battaglia, Pericic) to 5.6% on the Crotian mainland (Pericic, Barac). The Croatian islands of Brac, Hvar and Korcula have a somewhat lower frequency of around 4%, but that falls with measurement tolerance.
- I1-M253, the Germanic marker, is generally reported around 5% throughout the region, e.g. 6.2% in Montenegro, 5.3% in Belgrade, 5.7% (Battagla) and 5.1% (Pericic) with FYROM Albanians, 4.7% with Kosovo Albanians, 4.8% in Sarajevo. Unfortunately, Battagla and Pericic differ when it comes to Herzegovina and mainland Croatia. Battaglia has no I1 in Mostar, but 7.9% in Croatia (excluding Osijek), while Pericic has Herzegovina at 4.95% and mainland Croatia at just 2.8%. Both, however, point towards a drop in Northern & Central Bosnia (2.5% in Tuzla, 1.5% in Zenica). Rootsi et al. (2005) reports 5.3% for the Croatian mainland (312 samples) and 2.0% for Bosnians (91 samples). In summary, with the possible exception of Nothern & Central Bosnia, especially the Zenica area, the I2a Din frequency pattern does not seem to have influenced the rather constant I1 distribution across the area.
- R1a, generally regarded as Slavic marker, OTOH, falls pronouncedly from the 30-35% level as observed among Slovenes, the Croatian mainland and Macedonian Greeks. The R1a decline on the Balkans is generally correlated to more E-M78. Thus Kosovo and FYROM Albanians have the lowest frequencies of only some 2%. Belgrade's relatively low R1a share of 15,9% may as well still be related to a higher E-M78 share there..
Hovewer, in the I2a-Din core area, the negative correlation between R1a and E-M78 breaks. Here, R1a frequencies range as low as 8.7% on Brac and 12% in Mostar (Battaglia) and the Herzegowina (Periric), respectively, while E-M78 frequencies are also quite low (though not as low as on the Croatian mainland).
- R1b, generally seen as IE or Celitc marker, is affected most. While around 15%-18% among Croatians, Albanians and Greek Macedonians, and 11% in Belgrade and Tuzla, it drops to below 4% in Sarajevo (3.6%), the Herzegovina (3.6%) and Mostar (2.2%), Korcula (1.5%), Zenica (1.4%), and Hvar (1.1%).
I wonder how the I2a-Din pattern could have been brought about by Slavic expansion. More specifically:
- How could this expansion have strongly affected R1b, but left E-M78 and I1 untouched?
- Why would this expansion have led to such a strong concentration of I2a Din around Mostar, while reducing, instead of simultaneously enhancing, R1a in the area?
To me, it rather looks like a traditional co-existence of I2a-Din (more north-westwards) and E-M78 (more south-eastwards) that has been overformed by successive incursions of R1b (Celts), I1 (Goths, Heruli etc.) and Slavs (R1a), with the Goths and Heruli mostly sparing out Central Bosnia, while Celts hardly and Slavs only to a limited extent made it to Herzegovina and the adjacent Dalmatian coast.
That would point to Palaeolithic continuity, if there weren't the TMRCA and diversity issues ported out by Sparkey. I don't feel qualified to comment on the former. As concerns I2a-Din diversity, however, both Rootsi and Periric report it to be high in Bosnia and Herzegovina, respectively. In fact, both suggest diversity to be highest there, within a wide area of high diversity that comprises most of central-eastern Europe from the Czech Republic towards Western Ukraine (Rootsi even has the high diversity area stretching as far to the North-East as Estonia, but his analysis includes other I2 clades aside from I2a-Din). The following picture shows I2a-Din frequency distributions (A/C) and the corresponding variance surfaces (B/D) that Periric determined from own sampling in the Balkans (A/B) and by incorporating results from other studies across Europe (C/D):

In short, after reviewing available research, I think that I2a-Din has already been present in the Dinaric Alps before the Slavic expansion, probably already before Roman times. The question is just whether it originated there, or expanded from further north, maybe the Carpathians around the sources of Dniester and Tisza, sometimes in the Neolithic, the bronze or the iron age. Judged by the diversity maps, an arrival by sea looks quite unlikely.
Sources:
http://www.draganprimorac.com/wp-content/uploads/2010/05/Battaglia.pdf
http://mbe.oxfordjournals.org/content/22/10/1964.full.pdf (Periric)
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1181996/ (Rootsi)