Probably much less than 50%. I don't think we have ANY solid figures on this, but I think we can conjecture, based on other information we have. Probably a few unit commanders who came from middle class or minor aristocratic backgrounds, some desperate lower-class individuals who had nowhere else to go or were unwillingly forced into service (conscription). The greatest contingent would be the sons of former soldiers (who by law were required to enroll) but two factors heavily mitigate against the likeliness of mass native Romano British enlistment in this period, as they do against native Roman enlistment across the Empire throughout the fifth century.

The first factor: Roman military service was no longer considered an attractive career prospect for native citizens. For one thing, the likelihood of dying in battle was far higher in the 5th century than it had practically EVER been before. And even if you didn't die in battle, discipline had only grown more brutal over time, and you now stood nearly as much chance of being executed or brutally maimed for a minor camp offense. For another thing, despite the huge amounts of money the Empire threw at the army in general, during the 5th century soldier pay was mostly 'in kind,' or in the form of goods and equipment and food provided, not cash, and what cash there was directly paid to soldiers was not in stable currency like the solidus but in relatively worthless bronze pieces that couldn't buy anything but a cup of ale or a bite to eat. This situation would not be rectified until Justinian's time. In the meantime, from basically the late 4th through early 6th centuries Roman citizens would go to extreme lengths to try and avoid service, even to the extent of injuring themselves or becoming the proto-serf 'coloni' of large landowners who could shield them from being picked up in one of the mass conscription efforts the Romans routinely turned to during this period.

Secondly, and related to the first factor: all our evidence strongly suggests the Romans were recruiting heavily from beyond the border in the 5th century. The northern Europeans were already living hard lives without the comforts of Roman citizenship, and coin money was less crucial to them than regular meals, decent housing, sanitation, and decent medical care, things the average Roman already had access to but which Northern Europeans beyond the frontier could not find any more readily than they would in the Roman army. The Romans also paid non-Roman recruits somewhat less than citizen soldiers, so on the whole non-citizens were more desirable. Additionally, Roman recruiting standards placed a high premium on tall, strapping young men, and because they were less affected by the microbial disease ecology of the Roman world than its citizens (during the Roman period average height across the empire, even in Britain, has been shown to have dropped because the people living within the Empire--although benefiting from more regular meals and improved sanitation and greater average prosperity and security at the same time--were no subject to the combined disease ecology of the entire Mediterranean and Near Eastern zones, and as such their bodies were constantly battling a much higher microbial load than the Northern Europeans beyond the border) they tended to be bigger and stronger, physically, than the average Roman male, so when voluntary recruitment, and not conscription, was in effect, a Northern European competing for a spot in a unit against a Roman citizen had several likely advantages which led to him being enrolled more often than his citizen competitor.

As far as whether any Romano British soldiers may have remained in Briton when the administration withdrew, we similarly have no way of knowing. It is likely there were at least SOME native Roman soldiers and officers left in uniform when the administration withdrew, although the bulk of Roman Britains armed forces had already been transferred off the island by Constantine III for his attempted usurpation, and once the administration departed (or was ejected as some sources suggest) their pay would have stopped coming. We have some evidence that many of the highly Romanized Britons may have migrated to Brittany over the next few decades, apparently with the armed forces to contribute to flagging Roman efforts to retain hegemony over the increasingly patchwork situation in Gaul in the 540s-560s, so they may have gone over then, but it is at least possible that many stayed, as well, either returning to their hometowns and becoming part of local defenses or taking up with one of the local elites (potentially Aurelius Ambrosius) to try and maintain some semblence of Romanitas between the withdrawal of Roman imperial administration and the total collapse of Roman order outside of Italy in the 560s. We just don't know.