I appreciate your observation that the location of a unit's origin does not reflect their recruiting ground - this is certainly the case on Hadrian's Wall - but there is a lot of evidence suggesting that regardless of the origins of troops, the unit's gods tend to stay put with them, a look at the names of the gods turning up on Hadrian's Wall long after you expect the bulk of the soldiers to be local is testament to this.

I completely agree that a units Gods or Goddesses tended to stay with them, and wonder to what extent that was fortuitous and to what extent it was deliberately encouraged to foster a esprit du corps and how we could tell the difference.

Certainly the origin of the troops does not seem to have made that much of a difference - there are lots of Epona dedications and representations associated with troops who originally came from Lower Germany, for example, but few Eponas actually turn up in Lower Germany. The example of the Syrian soldier making an Epona dedication also demonstrates this.

Subsequent deployment into other units, or into civilian or administrative duties (beneficius consularis, exploratio, exercitor, etc) seems to have resulted in the retention of the original deities - perhaps they were a source of pride and also, evidence of close personal links withthe figure of the emperor (in the case of the eq. sing.)

This might also explain why there is no evidence for subsequent Epona worship once leaving the army and leaving the empire - as many soldiers from greater Germany did. They took back their training, their status, they were buried with swords and spurs, etc, but they did not raise roman-style altars or even have portable votive objects.

Which implies both an esprit du corps and an old school tie network or political framework in which to use it. Outside that framework, things were done differently.