Traditionally, you define ℚ as equivalence classes of pairs of integers. The integer 0 is an integer, it’s not an “equivalence class of pairs of integers” and therefore it’s not a rational, in the set-theoretic sense.
Real numbers have lots more constructions.
https://en.m.wikipedia.org/wiki/Construction_of_the_real_num...
I gotta figure out dedekind cuts or at least learn tarski's fixed point theorem which I'm sure that would allow me to much better understand tarski's construction
all in the slow lifelong process of understanding and learning to draw post's lattice
finally, to answer your question, I disagree with saying that there's funky business between ℤ ⊂ ℚ because of my alleged claim that there is at least one construction which avoids the problem you describe, but as I was trying to say, these would have a non-unique way to construct number zero which nobody likes
If you do it that way then there is only one version of “1”
I think the real funky business happens right between ℚ ⊂ ℝ
but maybe all I'm really trying to say is "hey, look at me, I understand how all of ℕ ⊂ ℤ ⊂ ℚ"
I'm very very sure that it's possible to give a fully ZF set-theoretic construction of the rationals using sets. in fact IMO the annoying thing is that there are in fact two trivially equivalent constructions that I can think of the top of my head, and I find that even worse than there being none at all