With a single particle, entanglement of course plays no role at all. Which is what Zeilinger says:
But Zeilinger uses it in his argumentation for his view. Look how he is doing it:
Zeilinger discusses 5 ways out after the confirmation that QM violates the Bell inequalities.
Deny realism
Deny locality
Deny counterfactual definiteness
Accept superdeterminism
Accept actions to the past
He more or less discards 3 4, and 5 rather briskly. Of locality, as already cited early, he remarks that most physicists think that we should give up on locality. However he tends to give up on realism, because this seems to be the conclusion of the Kochen-Specker theorem, and its first empirical tests. So the KS theorem has directly nothing to do with Bell states. But for Zeilinger it is a hint that of 'local realism' (which, as Joigus explained means 'locality' and 'realism' taken together), we have to loosen our concept of a reality behind our quantum measurements.
Please reread the chapter 'What could that mean?'. (Warning: he does not discuss these in the exact order as I did here. I streamlined his argument here. First he mentions the first three assumptions, then he argues against (3), then he discusses locality and realism, and only then he mentions superdeterminism, retro-temporal causation, just to discard them immediately.)
@joigus: as you see, Zeilinger argues against superdeterminism. He treats it as a kind of 'last straw'. This is his argument:
For completeness the 5th: