isoglosses and substrate

Hello,
I have a question regarding the so called "jat boundary" in Bulgarian
which is said to delineate the reflex of the Old Slavic jat'. To the
east of the boundary it is, for example, 'mljako', 'mjasto' (milk,
place), and to the west - 'mleko', 'mesto'. The jat boundary, which
also roughly coincides with many other isoglosses, runs from the
mouth of the river Isk&r, a tributary of the Danube, to the south
towards the Aegean coast along the rivers Struma/Strimon or
Mesta/Nestos.

It is thought that 'mljako', 'mjasto' were the original forms for
Eastern Balkan Slavic and that the 'e'-pronunciation is secondary.
However, it is said that the northern Danube tributary Olt, nearly
opposite to the Isk&r on the other side of the Danube, is also a
dialectal boundary for the Romanian language. Moreover, Vl. Georgiev
claims that the same boundary existed in pre-Slavic times – village
names to the west of Olt had been recorded as ending in -deva, and
others to the east - in -dava. Also, the modern river name Cib&r to
the south of the Danube had also been recorded as either Kebros or
Kiabros. What is the nature of this dialectal boundary in Romanian
and what are the current views on its origin there? Could it be
really due to a substrate or it is thought it has (re)established
itself with time over an initially homogeneous population? Is there
evidence from elsewhere of dialectal (?) differences in a substrate
affecting dialectal changes in a newly imposed language?

Regards,
Vassil