Short answer: physics and molecular orbital theory.
Long answer:
Pigment based colours work like this. All molecules have electrons, those electrons lie in their lowest-energy states, but they can be excited to higher energy levels by photons. If a photon bumps an electron up to a higher level, that photon of exactly that energy gets absorbed, but all other photons of all other energies pass through. You observe what’s left. If a pigment absorbs violet photons, you see the colour complementary to violet, which is yellow. If a pigment absorbs both violet and red photons, you’ll see the colours in between: OYGCB, which you’ll perceive as green.
For you to observe blue, the pigment needs to absorb orange, but no other colours in the higher-energy range of the visible spectrum. No yellow or green or cyan or violet absorption, or else the perceived colour is skewed. So it has to absorb one specific type of photon, and excite electrons by about 2 eV, but there cannot be any other energy levels above it: no other places to excite the electron at 2.2 or 2.4 or 2.5 or 3.0 eV, just the one excited state at 2eV and nothing else.
That’s very hard to do. Most molecular structures will give you lots of excited states that are relatively close together. It requires some unusually restrictive structural features in the molecule to limit the excited states like that. And it’s difficult for Nature to synthesize those molecules.
That’s not a problem for other colours. You want yellow? Fine. Absorb violet. And absorb everything above it, too, but that’s UV and humans can’t see it. You want green? Fine? Absorb red and also absorb violet, leave the green. Multiple excited states at 2.0 and 3.0 eV, no problem. But just absorb orange and nothing else? Tricky.
